The murine local lymph node assay is a predictive test for the identification of skin-sensitizing chemicals. The method has been the subject both of national inter-laboratory studies and of extensive comparisons with guinea pig tests. In the investigations reported here, the local lymph node assay has been evaluated further in the context of an international study comprising five independent laboratories. In addition, the influence of minor modifications to the standard assay procedure on the performance of the test has been examined. The modified procedures investigated were exposure of mice for 4 rather than 3 consecutive days, excision of lymph nodes 4 rather than 5 days after the initiation of exposure and the use of an alternative isotope. All five laboratories, irrespective of whether the standard or a modified protocol was used, were able to identify accurately, and with comparable sensitivity, potassium dichromate and 2,4-dinitrochlorobenzene as skin sensitizers. Using standard criteria, none of the laboratories recorded positive responses with methyl salicylate, a non-sensitizer. In the standard protocol, lymph nodes are pooled for each experimental group and the vigor of responses measured as a stimulation index relative to vehicle controls. A stimulation index of 3 or greater is considered to indicate skin-sensitizing potential. One further modification adopted by three of the laboratories was to analyze nodes from individual animals and, thereby, permit statistical evaluation. This allowed a direct comparison of statistical significance with the conventional stimulation index as criteria for a positive response. The data indicate that, while statistical evaluation may provide, in some instances, for small increases in sensitivity, this may be at the expense of some loss of selectivity. There are, however, insufficient data presently to draw firm conclusions regarding the relative value of statistical analysis. These studies demonstrate that the local lymph node assay is sufficiently robust to accommodate minor procedural and technical modifications without material changes in test performance.