One of the purposes of the amplitude correlation method was to investigate the possibility of reducing the number of similar events in our stress tensor inversion scheme . Events having very similar focal mechanisms do not contribute any independent information to the stress tensor inversion and since the inversion is rather slow we would like to reduce the number of events in each inversion as much as possible without affecting the result. Amplitude correlation is one possible way of performing such a reduction. Figure 8 shows the inversion results using all 100 events in the Ölfus dataset in the upper section of the figure. Below is the result obtained after removing 54 events considered similar by the amplitude correlation algorithm. There were 6 groups of similar events, with a total of 60 events in the groups. We retained the event with the largest number of stations and phases as the representative for each group. We see that there are more normal faulting stress states after the reduction but that the overall stress state is rather similar to the inversion with all events. We aim toward a better match between the stress results obtained after reduction with the original, we are in the process of investigating the use of composite fault plane solutions as representatives of the groups and a more accurate calculation of the confidence limits.