Degree-day (DD) estimates based on different numbers of daily temperat
ure readings recorded at regular intervals were analyzed for precision
and bias. Given a temperature threshold below the daily minimum, one
median standard deviation was 0.067 + 0.103, -0.043 (95% confidence li
mits) DD. Bias in the degree-day estimates was detected as a function
of time of year, and the relative position of the threshold between th
e minimum and maximum temperatures, however, the bias was less than 0.
030 + 0.00193 (SEM) DD per day. Hourly temperature data provided a bet
ter estimate of degree-days than mathematical functions fit to minimum
and maximum temperature data. Use of mathematical functions results i
n function specific bias that changes with time of year. In those inst
ances where the timing of phenological events has been based on a part
icular function, continued use of the function is necessary for consis
tent results.