Determining similarity among recipes is disclosed. A computer-implemented method of one embodiment of the invention receives a first recipe. The method determines one or more second recipes that are similar to the first recipe, and then outputs the one or more second recipes. The second recipes may be determined as being similar to the first recipe based on one or more different factors in one embodiment. Such factors may include: a weighted ingredient coefficient; a weighted same-type coefficient taking into account whether recipes are of the same type; a weighted same-title words coefficient taking into account the extent to which non-common title words of one recipe are also within the titles of other recipes; a weighted shared-keywords coefficient taking into account shared keywords among recipes; and, a weighted shared-ingredients coefficient taking into account the extent to which ingredients are shared among recipes.