Despite the importance of anecic earthworm burrows as preferential flow paths by which percolation water can bypass the rootzone, their spatial distribution has not been measured. The spatial distribution, i.e. whether worm burrows occur in aggregated or disperse distributions in a field, may affect the magnitude and variability of percolation in a field. We tested the following questions: (1) Are earthworm burrows distributed randomly, (2) is the variability and average of percolate in field plots with and without worms a function of burrow distribution. In a field experiment, we added 0, 25 and 50 worms/m2 to four field plots each. The plots were fitted with 300-cm2 zero-tension lysimeters at 40 cm depth before planting corn. The lysimeters were monitored during rain events. Only intense rain events in which rainfall rate exceeded infiltration rates yielded percolate, suggesting that ponding had to occur for macropore flow to begin. Earthworm burrows excavated by applied L. terrestris were randomly distributed and followed a Poisson distribution with a mean of the number of excavation mounds per 300-cm2 rectangles, i.e. the lysimeter size. We found that percolation volumes differed between plots with and without earthworms but that when considering the three treatments, no difference existed. This was mainly due to the high variability in the 25-worm treatment plots. We hypothesized that, for this treatment, the likelihood that earthworm burrows intersected the lysimeter pan was similar to the likelihood that they did not. Thus, variability was expected to be high, with some lysimeters acting like 50 worm lysimeters and others like 0 worm lysimeters. We tested this hypothesis using Inverse Monte Carlo simulation to simulate percolation volumes based on rainfall and earthworm distributions. The theoretical distributions thus obtained and measured distributions of percolate volumes were not statistically significant.