The paper evaluates the skill of precipitation forecasts from high-resolution numerical weather prediction (NWP) models for convective events. The Met Office Unified Model was run with grid spacings of 12, 4, and 1 km for 10 days in 2003 and 2004. The 1-km model was found to be the most skillful over all but the smallest scales (approximately <10–15 km). A measure of acceptable skill was defined, and the 1-km model achieved this at scales around 40–70 km, some 10–20 km less than the 12-km model. The biggest improvement occurred for heavier, more localized rain, despite it being more difficult to predict. The 4-km model did not improve much on the 12-km model because of the difficulties of representing convection at that resolution, which was accentuated by the spinup from 12-km fields. The verification method used a binary field approach, converting radar and forecast data into binary fields based on thresholds. Fractions were then calculated for different spatial scales, and skill scores were derived using mean squared error (MSE) and fractions skill score (FSS). The FSS showed that the 1-km model was more skillful at smaller scales, while the 12-km model was more skillful at larger scales. The 4-km model showed similar skill to the 12-km model at larger scales but was less skillful at smaller scales. The results showed that the 1-km model was more skillful for higher-accumulation thresholds, which have the biggest societal impact. The improvement was due to convection being explicitly represented rather than parameterized and due to a more accurate representation of predictable local effects. The 1-km model is still under development and has not been tuned for operational performance. The 4-km model performed poorly compared to the 1-km model and showed little improvement over the 12-km model at any scale apart from high-accumulation thresholds. The 4-km model had a longer spinup period and produced too large, intense, and well-spaced showers, leading to errors in the location and amount of rainfall.The paper evaluates the skill of precipitation forecasts from high-resolution numerical weather prediction (NWP) models for convective events. The Met Office Unified Model was run with grid spacings of 12, 4, and 1 km for 10 days in 2003 and 2004. The 1-km model was found to be the most skillful over all but the smallest scales (approximately <10–15 km). A measure of acceptable skill was defined, and the 1-km model achieved this at scales around 40–70 km, some 10–20 km less than the 12-km model. The biggest improvement occurred for heavier, more localized rain, despite it being more difficult to predict. The 4-km model did not improve much on the 12-km model because of the difficulties of representing convection at that resolution, which was accentuated by the spinup from 12-km fields. The verification method used a binary field approach, converting radar and forecast data into binary fields based on thresholds. Fractions were then calculated for different spatial scales, and skill scores were derived using mean squared error (MSE) and fractions skill score (FSS). The FSS showed that the 1-km model was more skillful at smaller scales, while the 12-km model was more skillful at larger scales. The 4-km model showed similar skill to the 12-km model at larger scales but was less skillful at smaller scales. The results showed that the 1-km model was more skillful for higher-accumulation thresholds, which have the biggest societal impact. The improvement was due to convection being explicitly represented rather than parameterized and due to a more accurate representation of predictable local effects. The 1-km model is still under development and has not been tuned for operational performance. The 4-km model performed poorly compared to the 1-km model and showed little improvement over the 12-km model at any scale apart from high-accumulation thresholds. The 4-km model had a longer spinup period and produced too large, intense, and well-spaced showers, leading to errors in the location and amount of rainfall.