The conventional method of determining excess reactivity and control-rod worth by summing reactivity decrements in a heterogeneous poisoning experiment may give rise to serious errors because of the neglect of interference effects between rods and poison. In this paper, it is suggested that interference may be accounted for by the simple assumption that it affects only the total control-rod worth but not the shape of the normalized reactivity versus height curve. It is shown that this assumption allows results to be extrapolated back to the rod worth in the unpoisoned core. The method is demonstrated in two poisoning sequences carried out in the Swiss swimming-pool reactor SAPHIR. The values obtained for excess reactivity and control-rod worth agree well with each other and with a direct measurement. It is shown that the normalized regulating curve is indeed independent of poison and that the method of summing reactivity decrements is seriously in error when applied to the second poisoning sequence.