Statistically, how many trials should you run before you can believe your results?

For a science experiment, how many times should one experiment be run before the data can be considered accurate, meaning all of the tests result in similar results and therefore, the result can be umed to be not a fluke.