Infrared attenuated total reflectance spectroscopy has been assessed for the analysis of extra virgin olive oil samples from various Mediterranean sites and their adulteration by sunflower oil. In this study two different silver halide fiber-optic probes were separately tested for the mid-infrared spectroscopic measurement of pure olive oil samples and these same oils adulterated with sunflower oil. One fiber-optic probe contained an exchangeable U-shaped section of the silver halide fiber, whereas the second probe consisted of a fiber-coupled diamond crystal, which performed slightly less well than the whole fiber probe. The optimum standard error of prediction for the sunflower oil concentrations in spiked olive oil samples, obtained by optimized partial least-squares (PLS) calibration models and leave-one-out cross-validation, was 1.2% by weight with the use of a special variable selection strategy based on a pairwise consideration of significant respective minima and maxima of the optimum PLS regression vector, calculated for broad spectral intervals. Calibration robustness was proven by also using packages of 10 randomly selected samples within a further cross-validation for calculating independent prediction values. The implications for product monitoring are discussed.