1: \begin{abstract} \noindent The complex Langevin method is a leading 

2: candidate for solving the so-called sign problem occurring in various 

3: physical situations. Its most vexing problem is that in some cases it 

4: produces `convergence to the wrong limit'. In the first part of the paper 

5: we go through the formal justification of the method, identify points at 

6: which it may fail and identify a necessary and sufficient criterion for 

7: correctness. This criterion would, however, require checking infinitely 

8: many identities, and therefore is somewhat academic. We propose instead a 

9: truncation to the check of a few identities; this still gives a necessary 

10: criterion, but a priori it is not clear whether it remains sufficient. In 

11: the second part we carry out a detailed study of two toy models: first we 

12: identify the reasons why in some cases the method fails, second we test 

13: the efficiency of the truncated criterion and find that it works perfectly 

14: at least in the toy models studied. 

15: \end{abstract}