Data compression is one of the important topics in computer science that is concerned in optimizing the size of data. There are many techniques for data compression used for specific type(s) of data and some others used for general purpose (i.e. for any type of data). In this work the lossless data compression is applied for iMulti media files (i.e. such as, image, audio, text). The implemented data compression algorithms are divided into three main classes: Run Length Encoding (RLE) techniques, Statistical techniques, and Dictionary techniques. RLE depends on removing the repetition of consecutive symbols. A slight modification is proposed for the counter of repetitions used in the standard RLE. Dictionary techniques try to replace context of symbols by small code. From this class LZW is studied in details and modified to produce mainly four other variants named LZWFA, LZWAL, LZWAU, and LZWIA. On the other hand Statistical techniques must build a statistical model first and then a small code is dedicated to the more probable symbols while longer ones are assigned to the less probable symbols. Also, the model can be of higher order than just one symbol that is predicting a context probability. Three different statistical models are proposed and implemented with Arithmetic coder. For each of the proposed approaches, an algorithm is constructed. These algorithms are tested by changing their parameters. A standard test files for given application are used in the tests. The Statistical Coder with order 3 model named PPM produced best compression ratio that exceeded the well known software WinZip(V.S) reduction ratio to about 3% on average and over 8% for best case. On other hand many modifications for the standard LZW were introduced. The best modification named LZWIA has improved the reduction ratio in the range of 7% on average to 27% under best conditions over the standard LZW reduction ratio.