Treeline research has strongly focused on mountain systems on the mainland. However, island treelines offer the opportunity to contribute to the global framework on treeline elevation due to their island-specific attributes such as isolation, small area, low species richness and relative youth. We hypothesize that, similar to the mainland, latitude-driven temperature variation is the most important determinant of island treeline elevation on a global scale. To test this hypothesis, we compared mainland with island treeline elevations, then we focused (1) on the global effects of latitude, (2) on the regional effects of island type (continental vs. oceanic islands) and (3) the local effects of several specific island characteristics (age, area, maximum island elevation, isolation and plant species richness). We collected a global dataset of islands (n = 86) by applying a stratified design using GoogleEarth and the Global Island Database. For each island we extracted data on latitude and local characteristics. Treeline elevation decreased from the mainland through continental to oceanic islands. Island treeline elevation followed a hump-shaped latitudinal distribution, which is fundamentally different from the mainland double-hump. Higher maximum island elevation generated higher treeline elevation and was found the best single predictor of island treeline elevation, even better than latitude. Lower island treeline elevation may be the result of a low mass elevation effect (MEE) influencing island climates and an increasingly impoverished species pool but also trade wind inversion-associated drought. The maximum island elevation effect possibly results from an increasing mass elevation effect (MEE) with increasing island elevation but also range shifts during climatic fluctuations and the summit syndrome (i.e. high wind speeds and poor soils in peak regions). Investigating islands in treeline research has enabled disentangling the global effect of latitude from regional and local effects and, at least for islands, a comprehensive quantification of the MEE.This article is protected by copyright. All rights reserved.