The purpose of ReCaptcha is to get the people filling it out to help identify unrecognisable words from books so that the books can be properly translated into a digital format. As numbers in books would generally be recognisable it is unlikely that they would ever appear in a ReCaptcha image. Generally a ReCaptcha displays one known word and one that they are trying to interpret. Entering the known word provides the CAPTCHA service while what everyone enters for the unknown word are compared to try to reach a consensus on what that word actually is.
If you don't care about helping to translate books to digital format then using an unobtrusive CAPTCHA - such as timing how long it takes to fill out the form - is the best option. You can always redisplay the form with a more obtrusive CAPTCHA for those who fill out the form too quickly and are therefore suspected of not being a real person.
If one of the fields in the form is a password then that field by itself should be the only CAPTCHA you need.