Prey selection by Somaniathelphusa sinensis (Parathelphusidae) among five gastropod species (two prosobranchs, three pulmonates) was associated with two snail attributes. First, the degree of defence of the tissues by the shell, approximated by shell weight (SW)/tissue dry weight (M). Snails with the heaviest shells (Sinotaia quadrata: Viviparidae) could not be attacked successfully by any crabs, and those with somewhat lighter shells (Melanoides tuberculata: Thiaridae) were protected from small crabs. Secondly, among those light‐shelled snails which could be attacked successfully, prey value (M/handling time, Ht) determined rank order of preference. Biomphalaria straminea (Planorbidae), with the highest M/Ht, was the species most readily eaten. When presented with a single snail species, crabs selected among different‐sized B. straminea in a way which maximized M/Ht. This preference was most strongly exhibited by small crabs. A lack of size selection for Radix plicatulus (Lymnaeidae) prey was in agreement with an absence of a clear influence of size on Ht for this species. Manipulation of M/Ht by attaching plastic discs to B. straminea shells (thereby increasing Ht) resulted in patterns of prey selection in accordance with the hypothesis that foraging S. sinensis maximized M/H1.Partial consumption of R. plicatulus, and to a lesser degree Physella acuta (Physidae), was common when groups of snails were presented to individual crabs, and occurred more frequently when large snails were offered. Partial consumption and subsequent rejection of large snails may have masked selection for small prey which were rarely abandoned before being completely eaten.The implications of patterns of prey selection were discussed with reference to the influence of crab predation on snail distribution and abundance in Hong Kong fresh waters.