The constancy of postmoult/premoult ratios of measures of linear size during ontogeny in insect and other arthropods is widely known as Dyar's rule. We tested this rule in nine species of the waterstrider genera Gerris and Aquarius (Heteroptera: Gerridae), using two size variables: head width and a multivariate measure derived from the pattern of multivariate allometry common to the species considered. Allometric patterns were similar in two independent datasets of laboratory-reared and field-caught specimens. Although our data strictly followed Dyar's rule injust a few instances, all growth ratios varied within a limited range only. Growth ratios for head width differed more between moults than those for multivariate size. The relationship between growth ratios for the two size measures conformed to the predictions based on allometry. We discuss hypotheses of the possible adaptive significance of growth ratios, such as their relation to mobility and systematic differences between hemimetabolous and holometabolous insects, and emphasize the importance of allometry. Since Dyar's rule is consistent with available evidence of physiological mechanisms underlying growth and moulting control of insects and crustaceans, it can be used as a general frame of reference to test alternative growth models.