The species-area relationship of insular faunas has been described and interpreted for over 70 years, but there is little agreement on its cause. Four processes have been proposed to explain this pattern: the equilibrium hypothesis, the passive sampling hypothesis, the habitat diversity hypothesis and the disturbance hypothesis. In this paper, we assessed the four hypotheses and analyzed the richness of island habitat avian communities and their influencing factors. Studies were conducted at 20 urban woodlots in Hangzhou, China during January to December in 1997. These urban woodlots are well separated from one another, forming isolated patches in urban matrix. Each woodlot was visited twice per month during bird migrant seasons in April, May, October and November, once in the other months. So each woodlot was visited 16 times all year. Twenty habitat variables, including woodlot area, shape, the degree of canopy, leave height diversity, horizontal heterogeneity, connectivity, disturbance level, surrounding, etc., were selected to enter multiple stepwise regression to test their influences on the richness of island habitat avian communities. 82 bird species were recorded at 20 urban woodlots all year. The fit equations of species number of one woodlot in one census (S) and species number of one woodlot all year (S_y) with woodlot area were: S = 2.7432A^0.3846 and S_y = 10.6574A^0.3669. At the basis of species-area relationship above, we discussed the value of z in the equation of S=CA^z and its significance, which has drawn much attention and remains questionable. z was traditionally thought to reflect the degree of isolation of an archipelago. We thought that z in the equation was just a constant and had no biological implication. It was affected by many factors besides isolation. To test the effect of sample size, species-area relationship and species density-area relationship of woodlot avian communities were compared. The results showed that species number increased with woodlot area increasing, while species density decreased with woodlot area increasing, which reveals that under the same sample size small woodlots have more species than large one. The best multiple regression models showed that species number of one woodlot in one census (S) was significantly affected by woodlot shape, tree species diversity, number of street tree strips connected to woodlot, leave height diversity, distance to the nearest woodland larger than 2 hm 2 and density of canopy layer. They explained 68.4% the variation in species number, among which tree species diversity was the most important factor influencing species number (58.3%). Species number of one woodlot all year (S_y) was significantly affected by woodlot shape and tree species diversity. The results of multiple regression indicate that besides the effect of sample size there are other factors more important influencing species richness. The avian community richness of Hangzhou urban woodlots were the results of synthesized action of multi-factors, including the effects of different sample size, the habitat diversity, disturbance factors, species factors and the effect of different research scales etc. Some of them significantly correlated with woodlot area. This is one of the causes that species number significantly correlate with woodlot area. Though the results of our studies partly supported the habitat diversity hypothesis and the passive sampling hypothesis, no of the four hypotheses alone could explain the species-area relationship of urban woodlot avian communities in Hangzhou. Natural communities are usually complicated. These hypotheses were raised on special communities and focused on some special habitat factors, so they couldn’t explain all communities. Most communities are affected by the synthesized actions of multi-factors. If the conclusion is considered as the synthesis hypothesis, the four hypotheses are the special case of the synthesis hypothesis. Models are the simplification of nature, but ecologists should be cautious to simplification. W...