We consider the effect of new physics on the branching ratio of Bs → l+l-γ where l = e, μ. If the new physics is of the form scalar/pseudoscalar, then it makes no contribution to Bs → l+l-γ, unlike in the case of Bs → l+l-, where it can potentially make a very large contribution. If the new physics is in the form of vector/axial-vector operators, then the present data on B → (K, K*) l+l- does not allow a large enhancement for B(Bs → l+l- γ). If the new physics is in the form of tensor/pseudotensor operators, then the data on B → (K, K*) l+l- gives no useful constraint but the data on B → K* γ does. Here again, a large enhancement of B(Bs → l+l-γ), much beyond the Standard Model expectation, is not possible. Hence, we conclude that the present data on b → s transitions allow a large boost in B(Bs → l+l-) but not in B(Bs → l+l-γ).