Channel catfish (lctalurus punctatus) were obtained from 3 lakes in southeastern Louisiana during the fall of 1986. A truss network of morphological measurements was developed, resulting in 10 variables measured on each individual fish. Morphological variables were adjusted to remove the effect of standard length through least-squares regression. Multivariate analysis of variance of adjusted variables yielded significant (P < 0.01) differences among lakes. Factor analysis of the partial correlation matrix of adjusted variables yielded 4 non-orthogonal dimensions descriptive of body shape variation, interpretable as: I) overall depth of body, 2) length of posterior portion of body, 3) caudal peduncle depth and location of anal and adipose fin insertions, and 4) location of dorsal fin insertion. Univariate analysis of factor scores showed significant (P < 0.05) differences among lake populations only on the first factor. Discrimant analysis based on adjusted variables correctly classified individual catfish to lake of origin at an overall rate of 84%. Results indicate that multivariate discrimination of catfish harvested from these lakes is suffi ciently accurate to facilitate formulation and enforcement of management regulations pertaining to these stocks.