The most widely adopted approach for estimating the wave-induced fatigue damage of the tubular joints of offshore structures relies on a wave scatter diagram to characterise the wave environment. This paper shows that the commonly adopted approach to compiling the wave scatter diagram and the subsequent selection of characteristic individual wave height and period pairs for analysis purposes will almost invariably lead to bias in the estimation of fatigue lives. A statistical argument is developed to show the sources and nature of this bias. This is supported by fatigue calculations based on individual wave data collected in the northern North Sea. It is found that the current practice in most cases introduces a substantial conservative bias into estimated fatigue lives. An alternative procedure for producing wave scatter diagrams and selecting wave heights and periods for subsequent finite-element analysis is presented which should lead to unbiased (or at least much less biased) estimates of fatigue lives. © 1996.