Cannabis sativa L. is an ancient cultivar that has gained much attention due to its versatility in different industries, e.g., medicine, textile and food. The quality control of the plant and their active pharmaceutical ingredients (APIs) is mandatory to guarantee the safety and effectiveness of the pharmaceutical product. In this context, Near Infrared Spectroscopy (NIRS) has emerged as a powerful technique due to its multiple advantages, e.g., non-destructive, cost-effective and rapid. In this article, a handheld NIRS has been employed, in combination with chemometrics, for the development of seven predictive models to determine five cannabinoids, along with moisture and nitrogen content in Cannabis samples, affording values of coefficient of determination of cross validation (R2CV) in the range of 0.75–0.99. To evaluate the suitability of the models, cross validation was performed, which provided low standard error of prediction (SEP) values, alongside slope values close to or equal to 1. Additionally, residual predictive deviation (RPD) was calculated for each parameter, obtaining values in the range of 1.67–9.36. Finally, the performance of the predictive results obtained has been compared with those achieved in previous research to demonstrate the suitability of the handheld instrument compared to two robust NIRS benchtop devices. Several improvements were highlighted for the determination of different parameters in Cannabis samples.