When results from a short-term experimental study of septic shock indicate that a therapy increases survival time (i.e., delays death), should the reader conclude that the therapy ultimately will increase the percentage of permanent survivors (i.e., prevent death)? If not, how long must animals be observed after lethal sepsis to reasonably validate that a therapy can be used to effect complete recovery? To help answer these questions, we reevaluated survival data from the symposium that entailed the largest number of separate shock therapy studies ever performed on a homogeneous group of dogs in one place at one time with a single lot of endotoxin. Twenty-eight scientists studied the responses of 115 treated and 38 untreated dogs given LD80 endotoxin by bolus injection to compare 15 therapies (most previously reported to have "increased survival"): the numbers of dogs alive were recorded at 0, 0.5, 1, 2, 4, and 7 days. Chi-square analysis indicated significantly more treated than untreated dogs alive only at 24 hr and significantly fewer treated dogs alive on day 7 than on day 2. Individual comparison of the proportion of animals alive in each of the 15 treated groups with the untreated group showed more dogs alive (P less than 0.05) in 3 treated groups at day 1 and in 1 treated group at day 2 and 4; so, four therapies appeared to "increase survival." However, by day 7 no therapy group had statistically more survivors than the untreated group. This evaluation indicates that in order to make a clear distinction between increasing survival time and increasing the percentage of permanent survivors, both treated and untreated dogs should be observed and compared after septic challenge until the percentage of survivors in both groups attains a steady state.