Compound eyes of insects in 16 orders were tested for the presence of a blood–retina barrier (BRB) by injecting the hemolymph with Procion yellow, which was excluded from the eye in all Neoptera but not in two apterygotes. A primitive apterygote (Petrobius, Machilidae) was investigated further. Epifluorescence observations with small dyes Lucifer yellow (LY) and sulforhodamine 101 (SR) confirmed uptake by the eye within 3 min of injection. LY and SR both penetrated the eye, particularly the cornea, examined in sections. Uptake was quantified by microfluorometry, yielding entry half‐times (t1/2) of 1–1.4 min, fitting predictions for a model where tracer uptake is limited by passive diffusion. A much larger fluorescent dextran entered at a similar rate (t1/2 = 1.70 ± 0.77 min; n = 22), too fast to be diffusion‐limited, pointing to an active process, probably flushing of hemolymph through the retina. This is not an artifact associated with tracer injection and may be the natural result of circulatory pressures. Microfluorometry gave a first estimate of hemolymph volume (2.9% of body weight), of hemolymph mixing time (t0.95 = 77 min); the eyes' receptive fields were also determined. All results point to a primitive crustacean‐like condition in Petrobius, with open access of hemolymph to the eye and no BRB. An evolutionary hypothesis is suggested to explain how a primitive central nervous system barrier later extended to cut off the eye in Neoptera, in the face of access problems for respiratory gases and metabolites. © 1999 John Wiley & Sons, Inc. J Neurobiol 41: 452–470, 1999