In the laboratory, under choice situation,C. longitarsus preferred shells with greater shell width and shell weight. The correlation coefficient values calculated between the crab and shell parameters in the laboratory sample showed that this hermit crab selects a shell of suitable dimension for its occupation. In the multiple regression equations calculated with carapace length against the three shell variables aperture width, shell width and shell weight which are deemed to be important for hermit crabs, the regression coefficient of the variables varied much. The unsealed first principal component explained 84·9% of the total variability of the shell parameters and scaled first principal component explained 95·8% of shell parameters. The scaled first principal component was found to be a reliable estimator of shell size and the hermit crab, if given a choice, selected a shell of suitable dimension which fits its body quite closely.