Day and night float counts for wood duck (Aix sponsa) broods were conducted on a 10.8 kilometer segment of the Holston River during the spring of 1973. Nighttime float counts were made using two boats, one down each side of the stream. Hand-held airplane landing lights were used to search the water's edge for roosting broods. Daytime float counts were made the morning following each night census on the same segment of river as the night census. The night and day methods were compared. Eight float counts during the day and eight during the night were completed. The mean number of broods seen at night was 29.5 (2.7 broods per kilometer). The mean number of broods seen during the day was 7.3 (0.7 broods per kilometer). There was a statistically significant difference between the day and night float techniques. The night counts were much more accurate and precise than the highly variable day counts. More than four times as many broods were seen at night as were seen during the day. Age classes and numbers per brood were more easily determined at night. It was concluded that the night float count technique was superior to the day float count method in determining numbers of wood duck broods on rivers.