Changes in chlorophyll a (chl a) and phaeopigment concentrations during 24 h incubations in water prefiltered through 2.0 pm Nuclepore filters were determined on a weekly basis over a period of 13 mo using water from Kaneohe Bay, a subtropical inlet in the Hawaiian Islands, USA. In bottles illuminated at a constant irradiance of 4.0 E m-' h-', both chl a and phaeopigment concentrations were consistently lower than initial values at the end of the incubations. Chl a concentrations declined at a lower rate in dark bottles than in light bottles. There was no evidence of a change in phaeopigment concentrations in dark bottles. There was no temporal pattern in the exponential decay rates of phaeopigments in light bottles over the course of the 13 mo study, the median value being 0.016 m2 E-' There was, however, evidence of a nonrandom temporal pattern in the chl a decay constants. Winter values were about twice as large as summer values, a result presumably reflecting changes in the physiology and/or species composition of the phytoplankton community. In about 30% of the incubations phaeopigment concentrations were higher than initial values at intermediate time points, in some cases by as much as a factor of 2 to 3 during the first 4 to 8 h of the incubations. These results are beheved to have been caused by stress associated with the effort to remove grazers by filtration. Inclusion of nanoplankton ( 2 to 10 pm) in the incubation bottles consistently resulted in a higher concentration of phaeopigments in the picoplankton fraction after 24 h than was the case in control bottles containing only picoplankton. In this study as in other work, prescreening through 10 pm filters appeared to be insufficient to eliminate grazing artifacts from phaeopigment photodegradation experiments. Dilution rather than filtration may be a more practical way to account for grazing effects in such studies.