A group of eighteen young chinook salmon, Oncorhynchus rschawyfscha Walbaum, trained to orientate in the direction 270", showed a unimodal orientation of 264" and a mean bimodal axis of orientation of 258"/078" (magnetic) within confidence limits of 218"+ +285": 038"-+105" when tested under controlled conditions in Auckland. Two years later 12 of these fish were transferred to a new test arena in Christchurch, south (7" latitude) of the original location, for individual re-testing at night in a light-proof room and also in sunlight; in the absence of ' local ' Auckland cues/clues. These fish had been kept in artificial light but six fish were exposed to sunlight in Auckland 2 weeks before moving to Christchurch.At night each fish showed consistent non-random orientation and nine out of 10 fish showed a mean bimodal orientation that fell within the confidence limit established in Auckland; the mean of means axis of orientation for 10 fish was 270"/090". In sunlight each individual fish showed a consistent, non-random mean bimodal orientation that fell outside the confidence limits established in Auckland and, at night, in Christchurch. The mean of means axis of orientation was 001c/1810 (magnetic); the map direction from Christchurch to Auckland is IS". The principal findings in this study were: (a) ' Local ' Auckland cues were not essential for location of the learnt direction. (b) A non-visual ' universal 'cue was used asa reference for location ofdirection at night. (c) Daylight orientation was influenced by anisotropic radiance, even though the sun image was obscured by the test arena cover. (d) The concentration of data at midday and sunset in artificial light and perhaps at midnight, may indicate the presence of a non-visual zeitgeber. (e) A correlation between time and direction, in light and in darkness, which by implication involves a co-ordination process. (f) The fish under test apparently imprinted on some aspect ofthe sun (in Auckland) within a 2-day period at the age of 2 years. (9) Displacement, by 7" in latitude, may have been detected by both visual and non-visual means.It is proposed that directional information from the geomagnetic field and the anisotropic radiance of sunlight provides a complementary cross reference which would be. sufficient for positional orientation, in relation to some fixed landmark, within coastal waters; but would not suffice for spatial orientation in mid-ocean.