||The mechanical alignment method is commonly used in many industries today because it is cheaper and easier to operate than other alignment methods. However, very little systematic research and study has been conducted in this area so far, and there are still some fundamental questions which remain unanswered. In this thesis, we investigated the precision level and established the baseline achieved by a mechanical alignment system using datums and reference surfaces. The factors which affect the accuracy of mechanical alignment system were studied and methodology was developed to suppress these factors so as to reach its full potential precision. In order to characterize the mechanical alignment system quantitatively, a new optical position monitoring system by using quadrant detectors has been developed in this thesis, it can monitor multi-dimensional degrees of mechanical workpieces in real time with high precision, which conventional measurement methods (such as interferometer) can not realize. We studied the noise factors inside the system and optimized the optical system. Based on the fact that one of the major limiting noise factors is the shifting of the laser beam, a noise cancellation technique has been developed successfully to suppress this noise, the feasibility of an ultra high resolution (<20 angstrom) for displacement monitoring has been demonstrated. Using the optical position monitoring system, repeatability experiment of the mechanical alignment system has been conducted on different kinds of samples including steel, aluminum, glass and plastics with the same size 100mmx130mm. The alignment accuracy was studied quantitatively rather than qualitatively before. In a controlled environment, we found that common practice of preparation of datum by insertion of precision machined rod into tightly fitted hole may not be adequate. The alignment precision can be improved 5 folds by securing the datum without other means of help. The roughness of reference surface of the samples can affect the alignment precision, the alignment accuracy of an aluminum workpiece having reference surface by milling is about 3 times better than by shearing. Also we have found that sample material can have fairly significant effect on the alignment precision of the system. Contamination trapped between the datum and reference surfaces in mechanical alignment system can cause errors of registration or reduce the level of manufacturing precision. In comparison with the results in a controlled environment above, we want to find out what will happen in the practice of industries where there is contamination and how much room there is in the system accuracy for improvement. In the thesis, artificial and natural dust particles were used to simulate the real situations and their effects on system precision have been investigated. In this experiment, we discovered two effective cleaning processes, one is the self-cleaning process, and another is the cleaning method by an air gun and a vacuum cleaner which is more efficient. We also studied how other factors (surface roughness, sample material) interacted with different types of contamination, and found that dust particles had less effect on the steel sample than on other samples.