Determining the causes of mortality in populations of fish is inherently difficult. To simplify the determination of whether parasite-induced mortality occurs, parasitologists have relied on 3 types of subjective analyses of graphs. Peaked host age-parasite intensity curves concomitant with a decrease in the degree of dispersion (measured by variance-to-mean ratio) of parasites in older age-classes of fishes, a slope of less than 2.0 for a log-log graph of variance versus mean intensity of infection, and differences between truncated and nontruncated forms of a theoretical frequency distribution for the parasite are considered indicators of parasite-induced mortality in fishes. The nematode Raphidascaris acus causes significant parasite-induced mortality in natural populations of yellow perch (Perca flavescens) in Dauphin Lake, Manitoba, Canada. Using this fish- parasite system we present a comparison of some of the graphical techniques used by parasitologists to detect parasite-induced mortality and show how confidence ellipses based on the parameters β0 and β1 of a linear model for growth of yellow perch (weight = β0 + β1 x age) can be used to compare many growth curves simultaneously. When plotted in a bivariate fashion (β0 vs. β1), vertical displacement of confidence ellipses along the ordinate (β1) are due to sublethal effects on growth of fishes in response to parasites, whereas lateral shifts along the abscissa (β0) are suggestive of parasite-induced mortality. Furthermore, this technique is robust even when intrinsic variation in the data is high, allows for the detection of other factors affecting growth of fishes, and could be applied to the comparison of different fish-parasite systems.