Two glass fiber samples, which had previously been tested for their ability to induce mesotheliomata in rats, were subjected to size fractionation and the respirable fibers from each were collected. The sizes of the fibers in the total and respirable fractions were measured and all four materials were tested in vitro for their cytotoxic activities against V79-4 cells, A549 cells, and mouse peritoneal macrophages. The respirable fraction from the coarser material had considerably enhanced activity (on a mass basis) compared with its parent (total) material. The respirable fraction of the other fiber was only slightly more active than the corresponding total sample. The fiber size distributions are discussed in relation to the observed biological activity of all four samples, and it is concluded that the cytotoxic activity of fibrous glass is determined by the number of fibers within certain size ranges. It is likely that in the systems used in this present paper, fibers less than 10 micrometer long are inactive. The relationship between these observations and those obtained with fibrous glass in vivo by Stanton et al. (1977) are discussed.