Tropical deforestation is one of the most important components of global change. Rates of deforestation in Brazil, the nation with the single largest concentration of tropical forest on Earth, have fluctuated widely over the last twenty years. Based on local knowledge, such fluctuations have been variously attributed to a wide range of factors such as the expansion of cattle ranching and soybean farming, infrastructural expansion and the proliferation of paved and unpaved roads, macroeconomic shocks to the Brazilian economy and international exchange rates. Many, if not all, of these arguments are plausible explanations for temporal variation in deforestation rates, but have to date not been subjected to rigorous statistical testing; this study investigates the potential impact of these variables on Brazilian tropical deforestation over the period 1990–2005. When analysed at the basin-wide scale, nearly all variables were highly inter-correlated through time and were also closely correlated with deforestation rate, but appropriate time-series analysis found no statistical evidence that any of the variables have systematically caused variation in deforestation rates. Power analysis showed that the variables may exert small or medium influences on deforestation rates, but the impacts, if present, are not strong. Future analyses of time series data at finer spatial scales that exploit spatiotemporal variation in deforestation rates and in the hypothesized predictor variables may find significant causal processes that are overlooked when analysed at the basin-wide scale.