Logic Programs with Annotated Disjunctions (LPADs) allow to express probabilistic information in logic programming. The semantics of an LPAD is given in terms of well founded models of the normal logic programs obtained by selecting one disjunct from each ground LPAD clause. The paper presents SLGAD resolution that computes the (conditional) probability of a ground query from an LPAD and is based on SLG resolution for normal logic programs. SLGAD is evaluated on classical benchmarks for well founded semantics inference algorithms, namely the stalemate game and the ancestor relation. SLGAD is compared with Cilog2 and SLDNFAD, an algorithm based on SLDNF, on the programs that are modularly acyclic. The results show that SLGAD deals correctly with cyclic programs and, even if it is more expensive than SLDNFAD on problems where SLDNFAD succeeds, is faster than Cilog2 when the query is true in an exponential number of instances.