Temporal patterns of stem and needle production and total aboveground net primary production (ANPP) were studied at the tree and stand level along four chronosequences of Siberian Scots pine (Pinus sylvestris L.) forests differing in site quality (poor lichen type or the more fertile Vaccinium type) and in frequency of surface fires (unburned, moderately burned (fire return interval of approximately 40 years), or heavily burned (fire return interval of approximately 25 years)). The maximum range of variability in aboveground production was quantified for: (1) possible long-term changes in site quality; (2) stand age; (3) non-stand-replacing, recurring surface fires; and (4) interannual climate variability. For (1) and (2), total ANPP was low in the lichen-type chronosequence, reached a maximum of 170 g C m(-2) year(-1) after 100 years and decreased to 100 g C m(-2) year(-1) in older stands. Maximum ANPP in the Vaccinium-type chronosequence was 340 g C m(-2) year(-1) and occurred earlier in the 53-year-old stand than in the other stands. Along the lichen-type chronosequences, peak ANPP was paralleled by maximum carbon allocation to stem growth. (3) In mature trees, damage by recurrent surface fires decreased stem growth by 17 +/- 19% over a 10-year period relative to pre-fire values. At longer timescales, ANPP was hardly affected by fire-related differences in mortality. (4) Needle- plus stem-NPP, reconstructed for a 3-year period, varied within a range of 15 g C m(-2) year(-1) in the lichen-type stands and 35 g C m(-2) year(-1) in the Vaccinium-type stands. For the same period, the coefficient of variance was higher for needle-NPP (20 +/- 10%) than for stem-NPP (12 +/- 7%). Needle- and stem-NPP did not covary in time. Most 30-year time series of stem-NPP at the tree level exhibited strong autocorrelation. In older trees, stem-NPP was positively correlated with growing season precipitation. Thus, the factors driving variability in ANPP ranked according to their maximum influence as: stand age (controlled by the frequency of stand-replacing fires) > site quality > growth depression because of surface fire damage approximately equal age-related reduction in ANPP > interannual variability approximately equal long-term effects of fire (stand density reduction). In lichen-type forests, we found that ANPP at the landscape level declined sharply when the interval between stand-replacing fires was less than 120 years, illustrating that fire strongly influences ANPP of boreal Scots pine forests.