Purpose: The purpose of this work is to study and quantify the differences in calculated dose computed with two algorithms available in treatment planning systems: Pencil Beam Convolution and Clarkson. Material and Methods: Four different types of treatment cases were analyzed: lung, head and neck, brain and prostate. For each case, the volume definition was based on a clinical CT-scan acquisition. The patients were treated with 3-dimensional radiation therapy. For each patient, 2 treatment plans were generated using exactly the same configuration of beams. In plan 1 and plan 2, the dose was calculated using the Clarkson and Pencil Beam Convolution algorithms, respectively, without heterogeneity correction. To evaluate the treatment plans, the monitor units, isodose curves, dose volume histograms and quality index were compared. A statistical analysis was carried out using Wilcoxon signed rank test. Results: The difference observed for monitor unites was 1.2% for lung and less than 1% for head and neck, brain and prostate. Wilcoxon test showed that there was “no statically significant difference, (p > 0.05)”. The dosimetric parameters derived from dose volume histograms were higher for organs at risks using Clarkson compared to Pencil Beam Convolution algorithm inviting clinician to make “safer” prescriptions. For quality index there was no statistically significant difference between both algorithms for all quality indexes, (p > 0.05). Conclusion: The clinical evaluation of a treatment plan should be made regarding the calculation algorithm which, in turn, is linked to the experience of the clinician.