Creme Brulee
A baked egg-yolk custard (cream, egg yolks, sugar, vanilla) set gently in a water bath, chilled, then topped with a layer of sugar caramelized to a brittle glassy crust just before serving.
Researched across 7 of 7 targeted sources.
The one thing
Everything about this dish exists to protect the yolks from ever getting hot enough to scramble: temper by pouring hot cream into the yolks slowly while whisking, then bake in a water bath only until the edges are set and the center still jiggles — the custard finishes setting as it cools. Skip either safeguard and you get grainy, curdled, or weeping custard instead of silky. s2 s4 s5 s6
Taste profile
Where it comes from
The earliest known printed recipe explicitly called crème brûlée appears in François Massialot's 1691 cookbook Cuisinier royal et bourgeois; but the underlying technique of browning a custard's surface with a hot iron predates this, documented as early as 1651 (La Varenne) and even in 1520's Libre del Coch for crema Catalana, and Massialot's own version placed already-caramelized sugar on top rather than torching a sugar layer as is done today. The dish's national origin is genuinely disputed: the English name 'burnt cream' comes from an English 1702 translation of Massialot, yet Massialot himself later called a similar dish 'crême à l'Angloise' in 1740, and crème brûlée was rare in French and English cookbooks until it surged in popularity in the 1980s, likely via Sirio Maccioni at Le Cirque in New York. s1 s3 s7
Crema Catalana
A related Catalan custard with a documented recipe from 1324 (Libre de Sent Sovi), often cited as older than crème brûlée; competing folk-origin stories trace it to Catalonia's Jewish community, Catalan convent nuns, Moorish introduction of cinnamon/lemon in 712, or English medieval custards. s3
Trinity Cream / Cambridge burnt cream
A version that spread to Trinity College, Cambridge in 1879, distinguished by the college arms being branded onto the custard surface. s1
What each ingredient is doing
The method, with reasons
Heat cream (with or without vanilla bean/flavorings) just to a simmer, then remove from heat
Bringing the cream just to the edge of boiling and then pulling it off heat raises its temperature enough to extract vanilla flavor compounds (if a bean is steeped) and to later shock the yolks into cooking evenly, without letting the cream itself scorch or reduce. s2 s4 s5 s6
Watch for: Cream shows small bubbles forming at the edges / a gentle simmer, not a full rolling boil
If you skip it: If the cream is allowed to boil hard or reduce, its fat and water ratio shifts and it can scald, giving the finished custard a cooked-milk flavor.
Steep vanilla bean pod and seeds in the hot cream for about 15 minutes, then straintechnique: infusion
The fat in the cream pulls flavor and aromatic oils out of the vanilla seeds and pod over time; straining afterward removes the solid pod fragments and seed husks so the custard stays smooth rather than gritty. s5 s6
Watch for: Cream turns fragrant and visibly speckled with vanilla seeds before straining
If you skip it: Skipping the steep (using paste or extract stirred in directly instead) gives a quicker, simpler vanilla flavor but with less depth than a true bean infusion.
Whisk egg yolks with sugar (and salt, if used) until combined and slightly pale
Whisking disperses the sugar throughout the yolks and begins to dissolve it, while also breaking up the yolks so they will heat evenly rather than in clumps once hot cream is introduced. s2 s5
Watch for: Mixture turns from deep yellow to a lighter, slightly pale, smooth consistency and the sugar grain disappears
Temper: add hot cream to the yolk/sugar mixture gradually, whisking constantly (sometimes with a small amount added first)technique: tempering
Adding hot liquid a little at a time while whisking raises the yolks' temperature incrementally instead of all at once. This lets the yolk proteins warm and begin to unfold gradually rather than seizing suddenly, which is what would happen if cold yolks were dumped into fully hot cream or vice versa. s2 s4 s5 s6
Watch for: Mixture stays smooth and liquid, with no small curdled flecks of cooked egg appearing
If you skip it: Adding the hot cream too fast, or all at once, cooks the yolks prematurely into small curds, giving the finished custard a grainy, scrambled-egg texture instead of a smooth gel.
Strain the combined custard through a fine mesh sieve
Straining removes any small bits of cooked egg, undissolved sugar, vanilla pod fragments, or foam bubbles from whisking, so the baked custard sets with a uniform, silky texture rather than trapped lumps or air pockets. s2 s6
Watch for: Liquid custard runs smooth and glossy through the sieve with only a small amount of solid residue left behind
If you skip it: Skipping this step (as some recipes do) risks small flecks or air bubbles surviving into the baked custard, marring its otherwise smooth surface and interior.
Bake ramekins in a hot water bath (bain-marie), water reaching partway up the sidestechnique: bain-marie
The surrounding hot water buffers and evens out the heat reaching the custard, keeping it below the temperature at which egg proteins would violently coagulate. Without the water bath, the direct oven heat at the ramekin's edges would run far hotter than the center, curdling the outer custard while the center stayed underset. s2 s4 s5 s6
Watch for: Water bath stays at a gentle simmer or below throughout baking rather than boiling vigorously
If you skip it: Baking without a water bath exposes the custard to direct dry oven heat, which tends to curdle the edges and create a grainy, weeping texture, and can form small holes or bubbles from overheated proteins releasing steam.
Bake until edges are set but centers remain jiggly/loose
The custard continues to firm up from residual heat as it cools after leaving the oven (carryover cooking). Pulling it while the center still wobbles avoids overshooting into a fully firm, grainy set. s2 s4 s5 s6
Watch for: When the pan is gently shaken, the custard's edges hold their shape while the center wobbles slightly like set gelatin, not liquid
If you skip it: Baking until the center looks fully firm in the oven means it will be overcooked once carryover heat finishes setting it, producing a rubbery, weeping custard instead of a silky one.
Cool, then chill the custard for a substantial period before serving
Chilling lets the custard's protein network finish firming into its final gelled texture and brings the dish to the cold temperature that creates the deliberate contrast against the hot, freshly caramelized sugar top. s2 s4 s5 s6
Watch for: Custard surface feels cold and fully set to a light touch, no longer warm or soft in the center
If you skip it: Topping and torching a still-warm custard risks melting or over-softening the base beneath the caramel and loses the hot-cold contrast that defines the dish at the table.
Sprinkle a thin, even layer of sugar over the chilled custard and caramelize with a torch or broilertechnique: caramelization
Concentrated, intense heat applied only at the surface melts and browns the sugar through caramelization before the underlying cold custard has time to absorb much heat, producing a thin, hard, glassy shell rather than a soft or dissolved topping. s2 s4 s5 s6
Watch for: Sugar turns from clear/white through amber to a deep golden-brown, and bubbles slightly at the surface before it is pulled from heat
If you skip it: Under-caramelizing leaves a grainy, undissolved sugar layer; overdoing it (especially under a broiler, which heats less precisely than a torch) can quickly burn the sugar to a bitter black rather than a deep golden brown.
Serve promptly after caramelizing (immediately or within a short window)
The caramelized sugar shell is a glassy, amorphous solid that is stable when dry and cool, but it slowly absorbs moisture from the custard beneath it over time, softening from a crisp shell into a sticky or dissolved layer. s4 s5 s6
Watch for: Tapping the top with a spoon produces an audible crack and the shell shatters into shards rather than bending or sticking
If you skip it: Letting the finished dish sit too long before serving allows moisture from the custard to soften the caramel, losing the crackling-shell effect that is central to the dish's appeal.
Can I riff on this?
Load-bearing
Egg yolks as the sole setting agent (no egg whites). Yolks alone give the soft, creamy gel that defines custard texture; whites would set too firmly and unevenly, turning the dish rubbery. s2 s4 s5 s6
Gradual tempering of hot cream into whisked yolks. This is the mechanism that prevents the yolks from curdling into cooked egg curds before baking even begins; every detailed recipe does this. s2 s4 s5 s6
Baking in a water bath with water partway up the ramekin sides. The water bath is what moderates oven heat enough for the custard to set gently instead of curdling; all detailed recipes use it despite varying the water depth. s2 s4 s5 s6
Stopping the bake while the center is still jiggly, not fully firm. Carryover heat finishes the set after baking; baking to a fully firm center in the oven overcooks the custard. s2 s4 s5 s6
Chilling the custard for a substantial period before serving. Chilling both finalizes the set and creates the cold base needed for the hot-cold contrast against the caramelized top; every recipe includes a chill window even though the specific hours vary widely. s2 s4 s5 s6
A separate sugar layer caramelized at the surface only, just before serving. This is the structural and sensory signature of the dish: a brittle, glassy hot crust over cold custard. Sources agree it is added and browned separately from the custard sugar, never baked in. s2 s4 s5 s6
Flexible
Dairy composition (all heavy cream vs. cream/half-and-half blend, and total volume). Changes richness and density of the custard but not the basic setting mechanism; sources differ here without disagreeing on method. s2 s4 s5 s6
Number of egg yolks relative to cream volume. Yolk ratio shifts richness and firmness of set but all versions remain a yolk-set custard; sources range from 4 to 6 yolks per roughly similar cream volumes. s2 s4 s5 s6
Oven temperature (325°F vs 350°F). Both are used successfully; 325°F is more common and gentler, 350°F bakes slightly faster with marginally higher risk of overcooked edges, but both rely on the same water-bath-moderated mechanism. s2 s4 s5 s6
Vanilla form and technique (whole bean steeped vs. paste or extract stirred in). Both deliver the defining vanilla flavor; steeping a real bean gives more complexity at the cost of time, while paste/extract is faster and simpler. Sources treat these as interchangeable alternatives. s2 s4 s5 s6
Additional flavorings such as salt or espresso powder. Salt and espresso are seasoning adjustments layered onto the same base custard; at least one recipe omits both and still produces the classic dish. s2 s4 s5 s6
Ramekin size/count and resulting yield. Portion size and shape (shallow wide vs. deep) change crust-to-custard ratio and cosmetic presentation, not the underlying custard chemistry. s2 s4 s5 s6
Exact chilling duration (1 hour to several days). All that's required mechanically is enough time to finish setting and chill through; the wide range across sources shows this is a scheduling choice, not a structural one. s2 s4 s5 s6
Caramelizing method (torch vs. broiler) and doneness check (visual vs. thermometer). Both heat sources achieve the same surface caramelization; a thermometer is offered by only one source as a more precise alternative to the shared jiggle test. s2 s4 s5 s6
Straining the finished custard mixture before baking. Only two of four recipes call for this extra smoothing step; it improves texture but is not universally treated as required. s2 s6
When it goes wrong
Make it yours — by goal
Make it faster / reduce total time
Skip whole vanilla bean infusion and use vanilla bean paste or extract stirred in directly instead of steeping a split bean for 15 minutes. Cuts roughly 15 minutes of steeping plus the extra straining step, since paste/extract can go straight into the hot cream or custard. Trade-off: Loses the more intense, complex vanilla flavor that comes from steeping real pod and seeds in the cream.
Chill for the minimum stated window (1-2 hours) rather than the longer make-ahead windows some recipes allow. Gets the custard to a servable, set texture sooner. Trade-off: Less time for the custard to fully firm and for flavors to meld compared to longer chills of 4 hours to multiple days used by other recipes.
Bake at the higher end of the temperature range (350°F) instead of 325°F. Shortens bake time somewhat since the oven runs hotter. Trade-off: Higher heat raises the risk of overcooking the edges before the center sets, versus the gentler, more curdle-resistant 325°F used by most other recipes.
Make it richer / denser custard
Increase the ratio of egg yolks to cream (toward 6 yolks per 2 cups rather than 4). Produces a denser, firmer-set, richer custard body since more yolk protein and fat are available to set the cream. Trade-off: Uses more egg yolks per batch and yields a heavier custard rather than the lighter set some recipes aim for.
Use all heavy cream rather than a half & half plus heavy cream blend. Increases overall fat content for a denser, richer mouthfeel and firmer set. Trade-off: Loses the lighter mouthfeel that the half & half blend option offers.
Make it lighter
Use the half & half plus heavy cream blend instead of all heavy cream. Reduces overall fat content for a lighter mouthfeel and slightly less rich custard. Trade-off: Custard sets slightly less firm and loses some of the dense richness that pure heavy cream provides.
Use the lower end of the yolk-to-cream ratio (4 yolks per 2 cups cream). Gives a slightly lighter-set custard rather than a dense one. Trade-off: Custard may set slightly softer/less firm than versions using more yolks relative to cream.
Add complexity or a flavor twist
Add optional espresso powder to the cream along with salt. Introduces a subtle coffee background note that layers with the vanilla and balances sweetness via the added salt. Trade-off: Moves away from the purely classic cream-and-vanilla flavor profile that recipes without extra flavoring maintain.
Steep a whole split vanilla bean (seeds and pod) in the hot cream for about 15 minutes, then strain, instead of just stirring in paste or extract. Extracts a more intense, complex vanilla flavor directly from the pod and seeds. Trade-off: Adds steeping time and an extra straining step compared to simply stirring in paste or extract.
More precise / reliable doneness
Use an instant-read thermometer to check for an internal temperature around 170°F in addition to the visual jiggle test. Gives a more precise, repeatable doneness signal than relying on visual judgment of jiggle alone. Trade-off: Requires an extra tool and piercing the custard surface, which the purely visual method avoids.
Better make-ahead flexibility
Chill the baked, unsweetened custards longer (up to 2-3 days) before adding and caramelizing the sugar topping. Allows the custard base to be fully prepared well ahead of serving, with only the quick torching step left for the day of. Trade-off: Still requires caramelizing and serving the sugar topping close to serving time, since the crackly crust does not hold well once made.