Dynamic imaging of Gd-DTPA uptake has been used by several groups to characterise the permeability of blood-brain barrier and blood-retina barrier lesions, using both bolus and constant infusion rate injections. However, no consensus on which injection protocol is most efficient has been reached. To address this problem, we extend our Simplified Early Enhancement (SEE) theory, applicable to retinal lesions, to cover infusion injections, and demonstrate its application to published data. The two injection methods are compared using computer simulation. We find that, first, an infusion cannot produce a constant plasma concentration in an acceptable time (although a hybrid injection, consisting of a combined bolus and infusion, is able to do this). Second, at any given time after the start of injection, a bolus achieves a higher tissue concentration, and hence enhancement, than does the same dose given as an infusion. Conversely, a bolus achieves any given tissue concentration in a shorter time than the same dose given as an infusion. Consequently, a bolus uses a smaller dose to achieve a given enhancement at a particular time. Third, if renal function is reduced, the error in calculating the permeability from a particular value of enhancement is lower for the bolus than for the infusion. And last, the SEE method is more accurate for a bolus than for an infusion. We conclude that a bolus is always more efficient than an infusion, as well as being easier to administer, and should always be used in preference to an infusion.