Whilst the RC reset is not the best solution these days, I would hazard to guess that you have other problems. This can be proven by using a reset chip as Steve suggested. Whilst I don't have first hand experience with the chip described, most devices have a resister that tells you what caused the reset - this will tell you if it was external or watchdog that caused the reset.

I would suggest that you rule out any obvious issues first:

1/ try another power supply - switchmode power supplies are normally fine for such an application.

2/ check your hardware construction - bypass caps etc. Don't expect a circuit wired together with little bits of wire to work.

3/ Is your crystal ok? Tap it with your finget and see if the circuit goes haywire. Sometimes crystals go 'flaky'.

4/ Is your software working correctly? Most problems are traced back to the device doing what it is supposed to, but not what you expected! The watchdog may be doing it's job properly - simplify your software in order to test this. A simulator may help you debug your software.

I wouldn't question if the chip is suitable for an industrial application, I would question whether your hardware and software are suitable. Think about it, the RC reset if very simple - if you power up you circuit - does the code run? If it does, the RC reset has done it's job. It's main downfall is that it doesn't reset on defined voltages, so it may not work with a 'brown out' condition. The reset chips measure the voltage and reset if the voltage is below the specified value. This is expecially critical for circuits that use eeprom - as corruption of the eeprom can occur if you try writing to it when the voltage is too low.