Here is his summary:
In summary, of six experiments testing a mental influence on machine-generated random events, four showed some evidence of PK. Two one-subject experiments showed significant psi-hitting when the subject was both relaxed and confident; a third one-subject experiment showed significant psi-missing when the subject was relaxed but not confident; a fourth experiment with ten unselected subjects showed marginal psi-hitting in an unfavorable environment; and the remaining two experiments were nonsignificant; one in an unfavorable environment, the other when the subject was doubtful of the outcome. All control studies were nonsignificant.
Note that the highly significant studies showed odds against chance of less than 1/100 and 1/2000, respectively.