Introduction: Atrial action potential duration (APD) alternans, a beat-to-beat alternation of long/short APD, may predispose to atrial fibrillation in patients [1]. Atrial APD alternans may result from a strong positive relationship between the APD and the stimulation basic cycle length (BCL), i.e. a steep APD restitution slope [2]; but some studies disagree [1]. Alternatively, APD alternans could result simply from a long steady-state APD measured at physiological rate (SS-APD), as studied in guinea-pig ventricle [3]. Aim: To test the hypothesis, for the first time to our knowledge in human atrial isolated myocytes, that longer SS-APD is associated with steeper APD restitution slope and/or larger magnitude of APD alternans. Methods: Custom-written software (WinEDR) was used to analyse action potentials previously recorded by whole-cell-patch clamp (35-37^oC) at various BCLs (between 150 and 1000 ms; 20 action potentials/BCL), in 50 human atrial myocytes. The myocytes had been enzymatically isolated from right atrial tissues obtained from 12 adult patients who were undergoing cardiac surgery. Procedures and experiments were approved by West of Scotland Research Ethics Service (REC: 17/WS/0134). Written, informed consent was obtained from all patients. The investigation conformed with the principles of the World Medical Association’s Declaration of Helsinki. All patients were in sinus rhythm on the day of surgery. All APDs were measured at a repolarisation level of -61 mV (i.e. APD_-61mV), to reflect the human atrial cellular effective refractory period [4]. Results: APD_-61mV restitution was fit to either a straight line (all 50 cells), or to a mono-exponential curve (achievable in 47 cells) for comparison. Dichotomising the SS-APD_-61mV values (measured at BCL 1000 ms), revealed a significantly steeper APD restitution slope to be associated with longer SS-APD: straight line slope =0.10±0.03 (mean±SEM; n=25 cells) vs -0.02±0.01 (n=25), P<0.001 (Mann-Whitney); maximum exponential slope =0.06±0.02 (n=23) vs -0.03±0.01 (n=24), P=<0.001. Furthermore, there was a significant and strong positive correlation (linear regression) between SS-APD and both straight line restitution (P<0.001, R²=0.577, n=50) and exponential restitution (P<0.001, R²=0.415, n=47). The maximum magnitude of APD alternans (calculated in each cell by averaging even-to-odd APD_{-61 mV}-subtractions of the last 5 action potential pairs at each BCL) was significantly larger in cells with longer (n=25) than shorter (n=25) SS-APD: 26±5 vs 10±2 ms, P<0.001. Furthermore, maximum APD alternans magnitude and SS-APD were significantly correlated (P<0.001, R²=0.390). However, there was only a weak association between APD alternans and APD restitution: steeper exponential restitution slope correlated significantly (P=0.021) but weakly (R²=0.113) with larger APD alternans magnitude; whereas there was no significant correlation between straight line restitution and APD alternans magnitude (P=0.316, R²=0.021). Conclusion: In human atrial myocytes, a longer SS-APD correlates both with a steeper APD-restitution slope and a larger APD alternans magnitude; but APD alternans magnitude is only weakly associated with APD restitution slope. Therefore, the propensity to APD alternans in human atrial myocytes, and, therefore, its potential to promote atrial fibrillation in patients, might be better predicted by a longer steady-state atrial APD measured at physiological rate, than by a steeper APD restitution slope.