Low-temperature growth limitation largely determines alpine treeline position globally, but treeline elevation also varies locally at a range of scales in response to multiple biotic and abiotic factors. In this study, we conceptualise how variability in treeline elevation is related to abiotic factors that act as thermal modifiers, physiological stressors, or disturbance agents. We then present a novel analytical framework for quantifying how abiotic factors influence treeline elevation at different spatial scales using New Zealand Nothofagus treelines as a case study. We delineated Nothofagus treelines in a GIS, along which we extracted data for treeline elevation and eight abiotic explanatory variables at 54 000 points. Each location was classified at each of five spatial scales based on nested river catchments, ranging from large regional to small hillslope catchments. We used hierarchical linear models to partition the variation in both treeline elevation and the eight abiotic variables by spatial scale, and then quantified the relationships between these at each spatial scale in turn. Nothofagus treeline elevation varied from 800–1740 m a.s.l. across New Zealand. Abiotic factors explained 82% of the variation in treeline elevation at the largest (regional) scale and 44–52% of variation at the four finer scales. Broad-scale variation in Nothofagus treeline elevation was strongly associated with thermal modifiers, consistent with the idea that treelines coincide with a temperature-driven, physiological limit. However, much of the finer-scale variation in treeline elevation was explained by a combination of thermal, physiological stress-related, and disturbance variables operating at different spatial scales. The conceptual model and analytical methods developed here provide a general framework for understanding treeline variation at different spatial scales.