A new optical fluorescence-based microsensor has been developed to continuously monitor intra-arterial blood gases (pH, pCO2, p02) with a probe sufficiently small to share the lumen of a 20 gauge arterial catheter without compromising pressure fidelity or ease of blood withdrawal. The evaluation of performance has proceeded through a combination of in-vitro, animal and clinical studies. Each has had its own unique problems. This series of studies has re-emphasized the fact that acceptable clinical performance of even the best designed systems is not assured by success in in-vitro or even animal studies. In-vitro studies were conducted to quantify the precision and time response of the sensors and to obtain an assessment of performance characteristics in the in-vitro environment. System precision (1 S.D.) in tonometered bovine blood was 0.03 (pH units), 2 mm Hg (pCO2), and 4 mm Hg (p02). These results were collected over 40 points at two different temperatures (28 and 37 F). Average drift measured over a 200 hour continuous study period showed 0.005 pH units, 0.9 mm Hg pCO2, and 2.1 mm Hg p02 per 24 hour period. In-vivo animal studies (dog, pig, sheep, and rabbit) demonstrated the need for antithrombogenic materials. The probe system employs a covalently bound Heparin coating. However, maintenance of a patent IV drip line and adequate arterial flow was found to be critical. With proper attention to these factors, overall probe performance was found to be equivalent to that predicted by the in-vitro studies. Clinical trials in volunteers, and in critical care and surgical patients, have re-affirmed the need for antithrombogenic probe materials and adequate arterial blood flow. In addition, human studies revealed factors which can lead to sensor offsets, unless properly compensated by probe design. Subsequent clinical trials in volunteers and clinical subjects have verified that continuous monitoring of blood gases is feasible with accuracies approaching that of the in-vitro blood gas analyzer.