Previously, we found a significant association between photosystem II efficiency (I broken vertical bar PSII) and photochemical reflectance index (PRI) measured at predawn among different species at different elevations and throughout several seasons. However, this relationship has not been evaluated under varied levels of illumination. Here, we used the Taiwan species Pinus taiwanensis (a conifer distributed at 750-3,000 m a.s.l.), Stranvaesia niitakayamensis (an evergreen tree, 1,700-3,100 m) and two Miscanthus spp. (perennial C-4 Gramineae, coastline-3,200 m) to elucidate the I broken vertical bar PSII-PRI relationship. We studied six levels of photosynthetic photon flux density (PPFD) (0, 200, 400, 800, 1,200 and 2,000 mu mol m(-2) s(-1)) over several growth seasons at high (2,600 m a.s.l.) and low (800 m a.s.l.) elevation sites. In comparing the same species or genus, I broken vertical bar PSII and PRI were closely correlated in darkness or under the same level of PPFD, with data obtained from different seasons and elevations pooled for regression analysis. Because both the intercept and slope of the I broken vertical bar PSII-PRI equation showed a negative curvilinear correlation with PPFD, we could fit an empirical regression model, I broken vertical bar PSII = c + d center dot ln(PPFD) + e center dot[ln(PPFD)](2) + f center dot PRI + g center dot PRI center dot ln(PPFD) + h center dot PRI center dot[ln(PPFD)](2), for multiple regression analysis. Using this model, we found a close correlation between the estimated and measured I broken vertical bar PSII (r (2) = 0.842-0.937, P < 0.001) for all four species examined and for mango (Mangifera indica) measured under both artificial illumination and sunlight (data from Weng et al. 2010). This empirical regression model could simulate both seasonal and diurnal variations of leaf-scale photosynthetic efficiency at high and low elevations