The data is in frequency form.
The observations are independent of each other.
Distribution basis is decided before data collection.
The sum of observed and expected frequencies are equal.
Sample size is adequate. In a 2x2 table, chi square should not be used when n is less than 20. In a larger table, no cell should have an expected value of less than 1 and not more than 20% of the cells can have expected values of less than 5.