ATP hydrolysis catalysed by the H+-ATPase of intact mitochondria can be induced by addition of ATP in the presence of valinomycin and KCl. This leads to an increase in intramitochondrial Pi and therefore allows investigation of potential Pi efflux pathways in intact mitochondria. Combining this approach with the direct measurement of both internal and external Pi, we have attempted to determine whether Pi efflux occurs via an atractyloside-sensitive transporter, by the classical operation of the Pi/H+ and Pi/dicarboxylate carriers, and/or by other mechanisms. Initial experiments re-examined the evidence that led to the current view that one efflux pathway for Pi is an atractyloside-sensitive ATP/ADP,0.5Pi transporter. No evidence was found in support of this efflux pathway. Rather, atractyloside-sensitivity of the low rate of Pi efflux observed in previous studies (oligomycin present) was accounted for by ATP entry on the well known ATP/ADP transport system followed by hydrolysis of ATP and subsequent Pi efflux. Thus, under these conditions, where ATP hydrolysis is not completely inhibited, Pi efflux becomes atractyloside sensitive most likely because this inhibitor blocks ATP entry, not because it directly inhibits Pi efflux. Substantial efflux of Pi from rat liver mitochondria is observed on generation of high levels of matrix Pi by ATP hydrolysis induced by valinomycin and K+ (oligomycin absent). A portion of this efflux can be inhibited by thiol-specific reagents at concentrations that normally inhibit the Pi/H+ and Pi/dicarboxylate carriers. However, a significant fraction of efflux continues even in the presence of p-chloromercuribenzoate, N-ethylmaleimide plus n-butylmalonate or mersalyl. The mersalyl-insensitive Pi efflux, which is also insensitive to carboxyatractyloside, is a saturable process, thus suggesting carrier mediation. During this efflux the mitochondrial inner membrane retains considerable impermeability to other low-molecular-weight anions (i.e., malate, 2-oxoglutarate). In conclusion, results presented here rule out an atractyloside-sensitive ATP/ADP,0.5Pi transport system as a mechanism for Pi efflux in rat liver mitochondria. Rather Pi efflux appears to occur on the classical Pi/H+ transport system as well as via a mersalyl-insensitive saturable process. The inhibitor-insensitive Pi efflux may occur on a portion of the Pi/H+ carrier molecules that exist in a state different from that normally catalysing Pi influx. Alternatively, a separate Pi efflux carrier may exist.