Adherent fibroblast-like cells from paired lines, one non-neoplastic and the other "spontaneously" transformed neoplastic, were compared in simultaneous in vivo and in vitro assays. The in vivo assay was the i.m. implantation of 10(6) or 10(7) cells in irradiated syngeneic animals, and the two in vitro assays were the evaluation of colony morphology on plastic and the enumeration of colony growth in semisolid agarose. The percentage of colonies diagnosed from their morphology as neoplastic correlated with tumorigenicity as follows: 100% always indicated a tumorigenic cell population with tumor latent periods from 6 to 230 days and tumor incidence from 40 to 100%; 0% always indicated a nontumorigenic cell population; 1 to 32% indicated either a tumorigenic cell line with long tumor latent period (218 days) with 70% tumor incidence or a nontumorigenic cell line. Growth in agarose, as measured by colony number and size, correlated with tumorigenicity as follows: nontumorigenic cell lines produced no colonies; tumorigenic cell lines produced colonies, but not always larger than 0.1 mm in diameter. The number of size or colonies did not correlate with the tumor latent period or tumor incidence. Therefore, both in vitro tests were reliable qualitative assays of spontaneous neoplastic transformation, but they did not correlate directly with the tumor incidence or mean tumor latent period. The relative success of the agarose assay emphasizes the importance of decreased anchorage dependence for progressive growth of injected cells as a malignant neoplasm in vivo.