Watershed restoration comes in a variety of forms depending on which set of problems are sought to be remedied. Severe soil erosion, in the form of gullying and/or headcutting, can be mitigated through constructing check dams in well-selected locations. This practice has been used throughout the upland subwatersheds within the Upper Laja River watershed in Guanajuato, México. The present study employed Wolman pebble counts to systematically assess the effectiveness of 21 check dams located near the city of San Miguel de Allende. Particle size distributions taken directly downstream and upstream of each check dam were differentiated, aggregated and compared-with the difference between median particle size of downstream and upstream distributions defined as DsD50-UsD50. Several subwatershed attributes were calculated in a GIS for comparison to DsD50-UsD50 values. Results indicate that, on aggregate, the check dams studied were moderately successful at retaining sediment that becomes entrained in concentrated flow (DsD50-UsD50 = 15.4 mm; p < 0.001). Individually, 18 of the 21 check dams surveyed had statistically significant differences between DsD50 and UsD50 (p < 0.05). The subwatershed variables of local channel slope (r = 0.55), mean subwatershed slope (r = 0.46), subwatershed area (r = 0.59), distance from channel head (r = 0.54), percent canopy cover (r = 0.46), Qmed (r = 0.46), total stream power (r = 0.58), and change in total stream power (r =-0.45) were found to be statistically significant when correlated with DsD50-UsD50 values (p < 0.05). Change in total stream power was used to classify stream reaches as either erosion or depositiondominated. When compared, the DsD50-UsD50 values from check dams located in erosion and deposition-dominated reaches are statistically different (p < 0.05); higher performing check dams were predominantly found in deposition-dominated reaches. The results of this ii study suggest that spatially distributed stream power can be used as a variable for making decisions about future check dam locations. In particular, check dams are more likely to be effective if they are located within deposition-dominated areas (negative change in total stream power) and in areas of decreasing change in total stream power. In general, this study's findings also support locating check dams in second order streams. iii Acknowledgments This work would not have been possible without the generous support of Agustin Madrigal and Mauricio Sánchez in San Miguel de Allende. I am also grateful to Ecosystem Sciences Foundation-Zach Hill, Derek Risso, and Tim Maguire-who facilitated the connection with Salvemos el Rio Laja and provided guidance and feedback when needed. Also, I would like to thank my committee members for their support and guidance particularly during the early stages of the project. Last, but far from least, I owe sincere thanks to my family, and particularly to my wife Jessie, for their endless encouragement.