This paper illustrates the usefulness of solvent control trials and presents a statistical analysis for mouse L5178Y lymphoma data. Examination of solvent control trials establishes that the natural logarithm of mutant frequency is approximately normally distributed and that both the mean and variance of mutant frequency vary by trial. There is little evidence of downturns at higher doses in the dose-response curves studied; therefore, a trend test is proposed for the detection of an increasing dose-response curve. A Monte Carlo investigation confirms that the proposed trend test is better able to detect an increasing dose-response than 4 alternate methods of analysis.