Numerous reports have concluded that zebrafish (Danio rerio) possesses A1-based visual pigments in their rod and cone photoreceptors. In the present study, we investigated the possibility that zebrafish have a paired visual pigment system. We measured the spectral absorption characteristics of photoreceptors from zebrafish maintained in different temperature regimes and those treated with exogenous thyroid hormone using CCD-based microspectrophotometry. Rods from fish housed at 15 degrees C and 28 degrees C were not significantly different, having lambda max values of 503 +/- 5 nm (n = 106) and 504 +/- 6 nm (n = 88), respectively. Thyroid hormone treatment (held at 28 degrees C), however, significantly shifted the lambda max of rods from 503 +/- 5 nm (n = 194) to 527 +/- 8 nm (n = 212). Cone photoreceptors in fish housed at 28 degrees C (without thyroid hormone treatment) had lambda max values of 361 +/- 3 nm (n = 2) for ultraviolet-, 411 +/- 5 nm (n = 18) for short-, 482 +/- 6 nm (n = 9) for medium-, and 565 +/- 10 nm (n = 14) for long-wavelength sensitive cones. Thyroid hormone treatment of fish held at 28 degrees C significantly shifted the lambda max of long-wavelength sensitive cones to 613 +/- 11 nm (n = 20), substantially beyond that of the lambda max of the longest possible A1-based visual pigment (approximately 580 nm). Thyroid hormone treatment produced smaller shifts of lambda max in other cone types and increased the half-band width. All shifts in photoreceptor lambda max values resulting from thyroid hormone treatment matched predictions for an A1- to A2-based visual pigment system. We therefore conclude that zebrafish possess a rhodopsin-porphyropsin interchange system that functions to spectrally tune rod and cone photoreceptors. We believe that these observations should be carefully considered during analysis of zebrafish spectral sensitivity.