We have examined the spatial-frequency selectivity of neurons in areas 17 and 18 of the adult pigmented ferret, by measuring how the amplitude of response depends on the spatial-frequency of moving sinusoidal gratings of optimal orientation and fixed contrast. Neurons in area 17 of the ferret respond optimally to low spatial frequencies [average 0.25 cycles per degree (c/deg)], much lower than the optima for cat area 17. The tuning curves are of the same form as those found in cat and monkey: unimodal with bandwidths in the range 0.8-3.5 octaves. Neurons in area 18 of the ferret respond optimally to even lower spatial frequencies (average 0.087 c/deg) than area 17 neurons, and the distributions of optimal spatial frequency for areas 17 and 18 hardly overlap. In both cortical areas, the bandwidth of the tuning curves is inversely correlated with optimal spatial frequency. This marked difference in tuning between the two cortical areas is probably attributable to differential geniculo-cortical projections. Small injections of fluorescent latex microspheres or horseradish peroxidase (HRP) were made into area 17 or area 18 in order to investigate the populations of geniculate neurons projecting to the two cortical areas. After injections into area 17, labelled neurons are found predominantly in the geniculate A layers, with a few neurons labelled in the C layers. Conversely, after an area 18 injection, similar numbers of labelled neurons are found in the C layers as in the A layers. Soma-size analysis of the neurons in the A-layers suggests the existence of two populations of relay neurons, which project differentially to areas 17 and 18. The different geniculate inputs and the different spatial-frequency tuning in areas 17 and 18 may imply that the two cortical areas process visual information more in parallel than in series.