As part of the Southern Oxidants Study, Brookhaven National Laboratory operated an intensive measurement site near Metter, Georgia, during parts of the summers of 1991 and 1992. Measurements were made of photochemically active trace gases and meteorological parameters relevant to determining causes for elevated ambient ozone concentration. The 1992 data set was used to calculate peroxy radical concentration and ozone formation rate based on determining the departure from the photostationary state (PSS) and based on a radical budget equation, such as applied previously to the 1991 data set. Averaged over the 28-day experimental period, we find maximum radical production occurring near noon at 2.5 ppb h-1, maximum peroxy radical concentration also occurring near noon at 80 ppt, and maximum ozone production of 8 ppb h-1 occurring near 1000 EST. Ozone photolysis accounts for 55% of radical production, HCHO and other carbonyl compounds about 40%. The radical budget and PSS methods depend in different ways on atmospheric photochemistry and a comparison between them affords a test of our understanding of the photochemical production of O3. We find that these methods agree to the extent expected based on uncertainty estimates. For the data set as a whole, the median estimate for fractional error in hourly average peroxy radical concentration determined from the radical budget method is approximately 30% and from the PSS method, 50%. Error estimates for the PSS method are highly variable, becoming infinite as peroxy radical concentration approaches zero. This behavior can be traced back to the difference form of the PSS equations. To conduct a meaningful comparison between the methods, the data set was segregated into subsets based on PSS uncertainty estimates. For the low-uncertainty subset, consisting of a third of the whole data set, we find that the ratio of peroxy radical concentration predicted from the PSS method to that predicted from the radical budget method to be 1.22¿32%. ¿ American Geophysical Union 1995