Micro-cylinders with the diameters spanning the 1-20 μm range are growing in importance, for realizing devices with new functionalities. For custom functionalities, their device-designs impose, tolerance-related constraints on their critical dimension. To meet the challenges for the associated online micro-metrology, new methods for the micro-cylinder diameter measurement, are currently receiving considerable attention. „Optical diffraction under Fraunhofer Approximation‟ is one of the most viable experimental techniques for cylinder diameter measurement, in the laboratory as well as Industrial environment. In 1-20 μm diameter range, however, the cylinder-diffraction is not well understood. The reliability of the current models/formulations for this range is far from satisfactory in respect of speed, accuracy, resolution etc., and need a re-examination. The present thesis concerns with a few investigations on the „Optical Diffraction by Micro-Cylinder‟. It highlights both the theoretical and the experimental aspects of the investigations on micro-cylinders with diameters in the range of 1-50 μm. The results of the investigation are organized into two categories. The first of them details a pair of new analytical models obtained from the principles of „Geometrical Theory of Micro-Cylinder Diffraction‟ while the second category highlights another pair of new analytical models obtained from the principles of the „Customary Fraunhofer Theory of Micro-Cylinder Diffraction‟. The model from the „geometrical theory‟ is based on the hypothesis that the ‘ray-paths relevant to the location of ‘diffraction minima’, facilitate to construct, a geometrically-equivalent triangle’. The solution of such a triangle provides the new formulation for the micro-cylinder diffraction. The model from the „customary Fraunhofer theory‟, instead, relies on the on the fact that „the diffraction pattern for a micro-cylinder is essentially, a chirped-interference pattern modulated by a diffraction envelope’. The functional form of the formulation depends upon, the type of triangle constructed for geometrical theory and the type of illumination used in the customary Fraunhofer theory. The thesis highlights, four new formulations (two from each of the approaches) to describe the micro-cylinder diffraction. The principal conclusions of the investigations are as under. - All the new formulations for the micro-cylinder diffraction facilitate, enhanced diameter inversion accuracy, in the hitherto esoteric diameter range of 1-20 μm. For the reported experimental data on 3 μm diameter micro-cylinder, the models proposed in the present investigation improve the accuracy of diameter-estimation from 16.5% known from earlier models to less than 1%. - The investigation also brings out for the first time, the hitherto unnoticed difference between slit-diffraction and the micro-cylinder diffraction: When the micro-cylinder diameter approaches the wavelength of the illumination, the first order diffraction angle approaches nearly 200. It may be noted that for a slit of same width, the corresponding diffraction angle approaches 900. When the critical dimension of the cylinder and the slit decreases from λ to 5λ, the difference in the corresponding diffraction angles reduce from nearly 700 to nearly 1.50. - The investigation also highlights that in the micro-cylinder diffraction for the said range of interest, the absolute intensity at the zero-order interference- maximum provides a new signature for the distance of separation between the diffraction minima. The consequence of this new finding is a considerable simplification in the apparatus and algorithm for diameter inversion from a diffraction experiment. The function of an array detector can be replaced by a point detector at a fixed point for all the diameters in the range of interest. - The two formulations proposed from the geometrical theory are suited for diffraction- minima search based algorithm, while those from customary Fraunhofer theory are well suited for intensity minima search based method for diameter inversion.