The coupling of optical fibers with Raman instrumentation has proven to be effective for real-time monitoring of chemical reactions and fermentations when combined with multivariate statistical data analysis. Raman spectroscopy is relatively fast, with little interference from the water peak present in fermentation media. Medical research has explored this technique for analysis of mammalian cultures for potential diagnosis of some cancers. Other organisms studied via this route include Escherichia coli, Saccharomyces cerevisiae, and some Bacillus sp., though very little work has been performed on Clostridium acetobutylicum cultures. C. acetobutylicum is a gram-positive anaerobic bacterium, which is highly sought after due to its ability to use a broad spectrum of substrates and produce useful byproducts through the well-known Acetone-Butanol-Ethanol (ABE) fermentation. In this work, real-time Raman data was acquired from C. acetobutylicum cultures grown on glucose. Samples were collected concurrently for comparative off-line product analysis. Partial-least squares (PLS) models were built both for agitated cultures and for static cultures from both datasets. Media components and metabolites monitored include glucose, butyric acid, acetic acid, and butanol. Models were cross-validated with independent datasets. Experiments with agitation were more favorable for modeling with goodness of fit (QY) values of 0.99 and goodness of prediction (Q2Y) values of 0.98. Static experiments did not model as well as agitated experiments. Raman results showed the static experiments were chaotic, especially during and shortly after manual sampling.