Light wood frame buildings are common in the United States, and due to their prevalence, there has been significant damage to this type of building in previous earthquakes. For this building type, shear walls (typically sheathed with oriented strand board) are the main lateral force-resisting element. Probabilistic models and fragility estimates incorporating the prevailing uncertainties for shear walls are needed to make accurate predictions of the effects of future earthquakes on buildings of this type. This paper uses experimental data to develop probabilistic capacity models for the shear force and drift of light wood frame shear walls at the yield, peak, and ultimate performance levels. The proposed models are applicable to shear walls in low-rise buildings where the shear deformation is dominant. Moreover, the proposed models are applicable to walls that are similar to walls in the dataset. The probabilistic models are constructed starting from deterministic models. For the shear force, a new deterministic model is proposed. This model predicts the shear force capacity of the wall based on the capacity of individual nails. For drift, existing deterministic drift limits are used. The deterministic models are then improved by incorporating the effect of additional physical characteristics through explanatory basis functions that are shown to be significant in a probabilistic model selection procedure. Finally, fragility estimates are constructed for an example shear wall using the proposed capacity models and conditioning on the shear force and drift demands.