This review article describes the problems involved in the calibration of a radiocarbon date (in conjunction with its associated error) to a calendar age. Recent investigations of the commonly quoted error of radiocarbon dates are discussed with their subsequent implications for calibration and currently available calibration schemes are critically compared. Results suggest that the incorporation of realistic error terms in any calibration scheme produces a considerable increase in the width of the typical calibration interval. Even when using a high precision scheme to calibrate normal precision dates, there is only a minimal narrowing of such intervals. A reduction in the width of the calibration intervals may be made by either calibrating a summary date obtained from replicate samples or, where possible, by calibrating a floating chronology.