The radiocarbon ages of dendrochronologically dated wood samples, each covering 10 years, are reported back to 2500 yr BC. The decadal calibration curve constructed from these data is an extension of the curve previously given for the AD interval (Stuiver, 1982). A major difference with the previous work, however, is the assessment of the error in the radiocarbon age determination. Whereas previously this error was only based on the Poisson counting statistics of the accumulated number of counts for the sample and standards, the current calibration error is based on an estimate of the reproducibility in the radiocarbon activity determination. As a consequence, the uncertainty in the current calibration curve is, on average, 1.6 times that of the AD curve previously given.