In this work we analyze a natural fractured carbonate reservoir in the Middle Magdalena Valley Basin (Colombia) using different volume curvature attributes. The main objective was to illuminate small or sub‐seismic faults, and to delineate fractured zones in the area. Most positive, most negative, dip, strike, curvedness, Gaussian, maximum, minimum and mean curvature attributes were calculated for a 3D post‐stack time migrated seismic volume. To reduce the noise level, a Butterworth filter was applied previous to the curvature attributes extraction and a combination of median and mean filters later. The most positive and most negative attributes gave the best results. They allowed identifying discontinuities at the target horizon, illuminating short wavelength flexures and lineaments. These features were associated to fracture zones that could result in hydrocarbon accumulation areas. Also, local geomorphological features and lineaments related to the structure axes were also identified using these attributes.