We compared transect counts used for the annual official count of male Kirtland's Warblers (Dendroica kirtlandii) to an observation-based mapping method of individually sighted males in 155 stands over 10 yrs. The annual census count almost tripled from 1990 to 1999. The transect and observation-based mapping method showed the same increasing trend in population between 1990 and 1999, except from 1992 to 1993. The annual official census transect count was consistently higher than the mapping method for stands censused in common. After standardizing for sample size, the annual number of Kirtland's Warblers per stand increased through time with the transect method, while there was a positive, non-significant trend with the mapping method. After 1992, the two methods began diverging in the number of males per stand. The relationship between the differences in count between methods (mapping count minus transect count) to the official transect count varied among years. At the stand level, the transect-method count was greater than the mapping count in 60% of the stands, while the mapping count was greater than the transect count in 16.3% of the stands. The difference in count between methods at the stand level ranged from −43 to 10. We illustrate the difficulties with interpreting transect counts due to Kirtland's Warblers' large territory sizes, occasional double territories, and active territorial defense during early morning hours. We suggest improvements to help correct these sources of error. We conclude that the official census transect counts are a satisfactory relative index, but results should not be interpreted as an absolute count.