Cell suspensions from 8-day chick embryo neural retina give rise to different populations in monolayer culture depending upon the combinations of substrata and media used. If a substratum of low adhesiveness is used, cell reaggregation takes place before attachment to the substratum, thus originating ''clumped'' cultures. In the presence of horse serum the clumps remain well contained and grow neuritic processes, but flat cells fail to appear. If fetal calf serum is present instead of horse serum, ''flat'' cells migrate out of the clumps and form a confluent monolayer. The clumps can be mechanically removed giving rise to a purified flat cell population. Clump formation can be prevented by seeding the cells at low density on highly adhesive substrata. In the resulting ''sparse'' cultures, flat cells do not develop even if fetal calf serum is present in the medium. Purified populations of process-bearing cells can thus be obtained on either polyornithine or highly adhesive collagen, with media containing either serum or the serum-free ''N1'' supplement. On polyornithine, however, cell survival in serum-free medium only takes place if the substratum is preexposed to serum before cell seeding. The cells present in the different monolayer cultures described above were further characterized using 3H-thymidine (3HT) incorporation as an indicator of mitotic activity, and tetanus toxin (TT) immunofluorescence as a neuronal ''marker''. The neuronal identity of the process-bearing cells present in clump-free, flat cell-free cultures was supported by the finding that 100% of them failed to incorporate thymidine and 95% or more were TT(+). In clump-containing cultures, TT(+) and 3HT(+) cells could be seen within the clumps with either horse or fetal calf serum. All of the flat cells which develop in the presence of fetal calf serum in these cultures appeared TT(–) and 3HT(+), a behaviour indicative of their non-neuronal nature.