A detailed characterisation of the performance of transmission Raman spectroscopy was performed from the standpoint of rapid quantitative analysis of pharmaceutical capsules using production relevant formulations comprising of active pharmaceutical ingredient (API) and 3 common pharmaceutical excipients. This research builds on our earlier studies that identified the unique benefits of transmission Raman spectroscopy compared to conventional Raman spectroscopy. These include the ability to provide bulk information of the content of capsules, thus avoiding the sub-sampling problem, and the suppression of interference from the capsule shell. This study demonstrates, for the first time, the technique's insensitivity to the amount of material held within the capsules. Different capsules sizes with different overall fill weights (100-400 mg) and capsule shell colours were assayed with a single calibration model developed using only one weight and size sample set (100 mg) to a relative error of typically <3%. The relative root mean square error of prediction of the concentration of API for the main sample set (nominal content 75%, w/w) was 1.5% with a 5s acquisition time. Models built using the same calibration set also predicted the 3 low level excipients with relative errors of 5-15%. The quantity of API was also predicted (with a relative error within ∼3%) using the same model for capsules prepared with different generations of API (i.e. API manufactured via different processes). The study provides further foundation blocks for the establishment of this emerging technique as a routine pharmaceutical analysis tool, capitalising on the inherently high chemical specificity of Raman spectroscopy and the non-invasive nature of the measurement. Ultimately, this technique has significant promise as a Process Analytical Technology (PAT) tool for online production application.