Climate models predict changes to the frequency and intensity of extreme events, with large effects on tropical forests likely. Predicting these impacts requires understanding how landscape configuration and land-use change influence the susceptibility of forests to disturbances such as wind, drought, and fire. This is important because most tropical forests are regenerating from anthropogenic disturbance, and are located in landscape mosaics of forest, agriculture, and other land use. This dissertation consists of four chapters that combine remote sensing and field data to examine causes and consequences of disturbance and land-use change in tropical second-growth forests. In Chapter 1, I use satellite data to identify factors associated with permanence of second-growth forest, and assess how estimates of carbon sequestration vary under different assumptions about second-growth forest permanence. I show that most second-growth forest is cleared when young, limiting carbon sequestration. In Chapter 2, I combine data from weather stations, remote sensing, and landowner surveys to model fire activity on 732 farms in the study area over ten years. The relative importance of these factors differs across scales and depending on the metric of fire activity being considered, illustrating how implications for fire prevention and mitigation can be different depending on the metric considered. Chapter 3 combines Landsat imagery and field data to map wind damage from a severe convective storm, providing strong empirical evidence that vulnerability to wind disturbance is elevated in tropical forest fragments. Finally, in Chapter 4 I integrate annual forest census data with LiDAR-derived topography metrics and tree functional traits in a hierarchical Bayesian modeling framework to explore how drought, topography, and neighborhood crowding affect tree growth, and how functional traits modulate those effects. The results from these studies demonstrate innovative approaches to understanding spatial variation in forest vulnerability to disturbance at multiple scales, and the results have implications for managing forests in a changing climate.