In this paper the attention was focused on the characterization of the slug flow generated by two fluids upstream (air–water) in serpentine micro-channels with squared sections and widths of {640, 320}μm. The curved geometries and the channels width greater than 100μm increase the complexity of the microfluidic process due to a weak presence of turbolence and inertial effect. All that produces the flow speed up and the enhancement of the mixing, but at the same time, a loss in the process control. The results presented are related to the changes in the slug flow pattern due to the different geometries varying the input flow rates and the investigated channel positions. A wide experimental campaign of a total of 69 experiments, divided in three experimental sets {set-1, set-2, set-3} was carried out.Two-phase microfluidic processes were monitored optically acquiring signals related to the light intensity variations in a selected channel position. Two flow patterns, slow and fast, were identified and the signals analysis procedure used has allowed their association to specific dynamical features both in time and in frequency domains. Four indicators were introduced for a quantitative evaluation of these features in the different operative conditions. The results have provided a characterization of the dynamics even though the nonlinearity of the process. In experimental set-1, comparing the flow patterns in two micro-channels with straight and serpentine geometries and width 320μm, it was proved the drastic changes induced in the flow displacement by the curves in terms of number of slugs and air/water presence. In experimental set-2, using a serpentine micro-channel with width 640μm, it was investigated the role of the input flow rate in the pattern formation and stabilization in terms bubbles length, frequency and inter-distance variability. Regarding to the experimental set-3, changing the investigated position in the micro-channel, the effects that the proximity to the inlet and the outlet leads in the segmented flow patterns were considered. From all the results, a great variability in the flows behaviours and a high sensitivity to the different operative conditions were evident, nevertheless, the wide potentialities in the use of serpentine geometries, for example to enhance the robustness to the input variations or to generate fast slug train at an established frequency, were proved. Moreover the low cost optical setting used integrated with the signal processing procedure could be easily adaptable to on-chip analysis for the processes investigations.