The reliability of an experiment is the consistency of the results. The degree of reliability is shown by the closeness of agreement of data for several repetitions of the experiment.
Reliability can be increased by taking care to make measurements carefully and accurately. Using measuring devices with the appropriate degree of precision will reduce the random error introduced by lack of precision. Repeating the procedure doesn't increase the reliability, but if you perform several trials and average the data the effect of outliers will be reduced. Repeating trials also allows you to see random errors and data points that can be disregarded
Validity is the degree to which the procedure tests what it's designed to test. Data can be reliable without being accurate or valid, for example if it's consistently inaccurate, but in order for data to be valid it must be reliable. Validity is increased by having controlled variables and using appropriate methods of measurement.