Blocking or synchronizing activity during regeneration of the retinotectal projection prevents both the sharpening of the retinotopic map recorded on tectum and the refinement of the structure of individual arbors within the plane of the map, and this refinement is triggered by N-methyl-D-aspartate (NMDA) receptors. We tested whether activity-driven refinement also occurs during development of the projection in larval and young adult goldfish. Shortly after hatching, larval goldfish were placed into tanks within light-tight chambers illuminated by a xenon strobe at 1 Hz for 14 h of each daily cycle. Fish were reared for 1.5-2 years, until large enough to record in our retinotectal mapping apparatus (6 cm length). Age- and size-matched controls had normal maps with multiunit receptive fields (MURFs) recorded at each tectal point of 10.8 degrees (0.16 S.E.M., n = 5), whereas the strobe-reared fish had only roughly retinotopic maps with much enlarged MURFs averaging 26.7 degrees (1.41 S.E.M., n = 5). This enlargement represents an abnormal convergence onto each tectal point, as the maps failed to sharpen during development. The arbors of individual retinal axons were stained with horseradish peroxidase (HRP) in larval fish and in adult strobe-reared and control fish. They were drawn with camera lucida from tectal whole mounts, and analyzed for spatial extent in the plane of the retinotopic map, order of branching, number of branch endings, depth of termination, and caliber of the parent axon. Arbors from larval fish (1-2 weeks) were small (approximately 50 x 40 microns) with less than 10 branches, occupied a single strata, and could not be separated into different classes by caliber of axon. The 87 arbors stained in control adult fish (6 cm long) were much like previously examined adult arbors, with those from fine, medium, and coarse axons averaging 115, 166, and 194 microns in extent, respectively, and having 17-24 branch endings. The 110 arbors from 12 strobe-reared fish were often abnormal. Although the fasciculation was normal, the extrafascicular routes were abnormal with reversing turns. The axons often had branches along their course, and these branches were scattered across a wider extent, rather than forming a distinct cluster. In contrast, neither the number of branches nor the depths of termination was significantly changed in any group. The coarse caliber arbors were most abnormal, being 64% longer and 30% wider than controls. The fine caliber arbors were also significantly larger by about 20%, but the medium caliber arbors were not enlarged.(ABSTRACT TRUNCATED AT 400 WORDS)