Using the mean $\mu = n \times p$ ($\text{sample size} \times \text{probability}$), compare it to the observed value. If it is more, than test upper, otherwise lower.

Note: The critical region is the closest value to passing the significance level, but not passing it.