Body volume was measured directly by the water displacement method and using three different geometric models, consisting of a series of cones and truncated cones, in California sea lion (Zalophus californianus) pups from the Gulf of California. Volume is required for the estimation of density, an indicator of the ratio of fat to lean mass. A model consisting of two cones and four truncated cones predicted body volume with greatest accuracy (volume=1·11+1·06&;estimated volume, r2=0·90, P<0·001, SEE=0·673, N=274). This model, however, was not adequate to estimate absolute pup body density (mass to volume ratio), as it yielded values poorly correlated with observed body density (r=0·14, P=0·02). However, the regression line of mass on volume indicated that density decreased with volume, so the mass to volume ratio is a biased estimate of density and is not appropriate for comparing whole body density among sea lion pups. The direct analysis of the relationship between body mass and body volume (observed or estimated) through analysis of covariance provided a better tool to compare the relative density among pups of different sex, populations, or born in different years. The results from such analyses are consistent with previous evidence of sex and age effects on body composition.