Five in vitro assays have been applied to screen the efficacy of potential chemopreventive agents. These assays measure a) inhibition of morphological transformation in rat tracheal epithelial (RTE) cells, b) inhibition of anchorage independence in human lung tumor (A427) cells, c) inhibition of hyperplastic alveolar nodule formation in mouse mammary organ cultures (MMOC), d) inhibition of anchorage independence in mouse JB6 epidermal cells, and e) the inhibition of calcium tolerance in human foreskin epithelial cells. The efficacy of many of these same agents in whole animal studies of lung, colon, mammary gland, skin, and urinary bladder carcinogenesis has also been measured. The aim herein is to estimate the positive and negative predictive values of these in vitro assays against whole animal chemopreventive efficacy data using the same chemicals. For three of these assays--using RTE, A427 cells and mouse mammary organ culture (MMOC)-enough data are available to allow the estimate to be made. Such extrapolations of in vitro data to the in vivo situation are difficult at best. There are many dissimilarities between the two assay systems. The in vitro assays use respiratory and mammary epithelial cells, while the in vivo assays use respiratory, mammary, colon, bladder and skin cells. The in vitro assays use the carcinogens benzo(a)pyrene (B(a)P) and 7,12-dimethylbenz(a)anthracene (DMBA), while the in vivo assays use B(a)P, DMBA, N-methyl-N-nitrosourea (MNU), N,N'-diethylnitrosamine (DEN), azoxymethane (AOM), and N-butyl-N-(4-hydroxybutyl)nitrosoamine (OH-BBN). There are vast differences in pharmacodynamics and pharmacokinetics in vitro and in vivo, yet it is possible to rapidly screen chemicals in vitro for efficacy at one-tenth the cost and complete tests in weeks instead of months. A positive in vitro assay was defined as a 20% inhibition (compared with control) for the RTE and A427 assays and a 60% inhibition for the MMOC assay at nontoxic concentrations. For in vivo assays, the criterion for a positive result was a statistically significant inhibition of incidence, multiplicity or a significant increase in latency (mean time to first tumor). For an agent to be considered negative in animals, it required negative results in at least two different organ systems and no positive results. Using the battery of three in vitro tests, the positive predictive value for having one, two, or three positive in vitro assays and at least one positive whole animal test was 76%, 80%, and 83% respectively. The negative predictive values for one, two or all three in vitro assays was 25%, 27%, and 50%. From these data it is observed that in vitro assays give valuable positive predictive values and less valuable negative predictive values. The mechanisms of chemoprevention are not well understood. Seven categories of agents were examined for their cancer preventing both in vitro and in vivo: antiinflammatories, antioxidants, arachadonic acid metabolism inhibitors, GSH inducers, GST inducers, ODC inhibitors, and PKC inhibitors. Three or even five in vitro assays cannot be all-inclusive of the many mechanisms of cancer prevention. However, three assays help to predict whole animal efficacy with reasonable positive predictive values. Much work and development remains to be done to rapidly identify new chemopreventive drugs.