The apparent discrepancy is analysed between the literature value of the level splitting calculated by ZINDO method for sexithiophene clusters and the Davydov splitting calculated for the three-dimensional crystal of alpha -sexithiophene in the dipole approximation with the transition moment distributed over the six rings. Consistency is restored by identifying the relevant interactions that are not taken into account in the cluster approach. The results suggest that the band in the alpha -sexithiophene absorption spectrum, previously assigned to the upper Davydov component, might be due to charge-transfer (CT) states; other evidence pointing to the same conclusion is reviewed.