AIMS: Improved boiling water target designs and body materials for the production of ( 18 F)fluoride ion have evolved steadily over the last two decades (1), and reliable operation at beam power up to one kilowatt has been achieved. Our goal is to extend this beam power limit to two kW for boiling water targets through use of the thermosyphon principle (2), and to ten kW or more through use of a recirculating system (3) utilizing a miniature regenerative turbine pump and a high efficiency low-volume heat exchanger. Preliminary experimental work on these approaches was reported at WTTC-9. We envisage the thermosyphon target as a possible retrofit for many existing PET accelerators if full use of the maximum available beam current is desired, and the recirculating target as potentially useful in accelerators with an injection ion source permitting extracted beam currents from 100 to 1000 microamps. METHODS: Prototype #4 thermosyphon (4) is being tested at the Duke PET Facility at 40 microamps of 22 MeV protons using the CS-30 positive ion cyclotron (deflector limited). We plan to optimize yield for this target for daily FDG production, and then design and beta test targets up to their performance limit at accelerators with beam power above one kW. For the recirculating target (5), we have constructed and tested prototype #2 turbine pump and prototype #1 heat exchanger, and are now fabricating a target body. These three components will be combined and tested as a system prior to running beam tests at 27 Mev and 40 microamps. LabView software and field point modules are used for the control systems of both targets. RESULTS: Prototype thermosyphon target beam tests have characterized operating parameters, and we are developing a predictive thermohydraulic model (6,7) for condenser and cooling channel geometry, plus using the MCNPX code to assess bubble void fraction/distribution. This model will facilitate high power designs. Recirculating target experiments have included separately characterizing operating parameters of the turbine pump and heat exchanger. A target body now under construction will be added to test the entire system, prior to Duke beam planned for late 2004. Predictive thermohydraulic models (8) are being developed which will be correlated with the Duke beam testing, permitting design of a compact integrated system with high reliability and a lower enriched water inventory for beta test at 2-10 kW accelerators. CONCLUSION: The goal of this work is more efficient use of PET accelerators through better targetry, leading to reduction in the cost of producing ( 18 F)radiopharmaceuticals.