Can a nonextremal black hole be a particle accelerator?

We consider particle collisions in the background of a nonextremal black hole. Two particles fall from infinity, particle 1 is fine-tuned (critical), collision occurs in its turning point. The first example is the Reissner-Nordstr\"{o}m (RN) one. If the energy at infinity $E_{1}$ is big enough, the turning point is close to the horizon. Then, we derive a simple formula according to which $E_{c.m.}\sim E_{1}\kappa ^{-1/2}$, where $\kappa$ is a surface gravity. Thus significant growth of $E_{c.m.}$ is possible if (i) particle 1 is ultrarelativistic (if both particles are ultrarelativistic, this gives no gain as compared to collisions in flat space-time), (ii) a black hole is near-extremal (small $\kappa$). In the scenario of multiple collisions the energy $E_{c.m.}$ is finite in each individual collision. However, it can grow in subsequent collisions, provided new near-critical particles are heavy enough. For neutral rotating black holes, in case (i) a turning point remains far from the horizon but large $E_{c.m.}$ is still possible. Case (ii) is similar to that for collisions in the RN metric. We develop a general theoretical scheme, direct astrophysical applications can be a next step to be studied.