There is a 90% probability of making a Type I error

There is a 90% probability of correctly rejecting the null hypothesis when it is actually false

There is a 90% probability of failing to reject a false null hypothesis

The effect size is 90% of the predicted difference