Considering seismological data sets as complex networks is an approach that can be used to study the seismicity in a region; comprehend the spatial and temporal complexity of this phenomenon; and even more important, unravel the relationship among seismic events. In this work, we consider 4985 events, selected from the U.S. Geological Survey (USGS) earthquake catalog for the Colombian region, to build earthquake networks based on the temporal sequence of events and their specific location. We show that the networks are of small-world type independent of the cell size, or granularity, used to construct the networks, which is similar to the results reported for other regions even though the Colombian region is a complex seismic region where many plates converge. Furthermore, this work characterizes the structure of the complex networks of Colombian seismicity and highlights zones with seismic importance such as the Bucaramanga sector. Based on these results, we could apply complex network techniques to test the possibility of forecasting seismic events based on their spatial and temporal sequences.