Poly-methyl-methacrylate (PMMA) microchannel based devices are gaining popularity, particularly in medical applications. Experiments were carried out on a PMMA based microchannel profile of thickness 10 mm through a 100-Watt CO2 laser machine by varying laser power, scanning speed and number of passes of the laser machine. The full factorial design of the experiment was implemented to investigate the microchannel profile, depth and surface roughness of the microchannel. The analytical model was developed to predict the microchannel profile and depth for the prescribed number of laser passes. The developed analytical model has been compared with the experimental investigations and also with the previously developed models available in the literature. Results of the analytical model predicted the fabricated profile with appreciable accuracy. Moreover, the close agreement of the results with the available literature models validates the acceptability of the proposed analytical model. The surface roughness predictive modeling using linear regression, polynomial regression of order 2,3 and 4, exponential Box-Lucas method and sigmoid Boltzmann models have been compared. The statistical analysis of all the models was performed, with the analysis of variance, mean squared error, Root Mean Square Error, mean absolute percentage error and coefficient of determination. Based on the comparative assessment, the best fitted and accurate model obtained was polynomial regression of order 3. Thus, the improved surface roughness generated with the laser machining would be a time and cost technique for the fabrication of microfluidic devices.