For a scope with a 1st order low-pass filter (most low to mid range scopes), risetime and bandwidth are interchangable: f_3 = 0.35 / t_r. 0.35 / (17.5 ns) = 20 MHz. There is no simple formula for a higher order filter, but usually it is slower, or the risetime is hard to define due to ringing.
If you see a rise time of 17.5 nanoseconds on a scope with a rated bandwidth of 5 MHz, one of three things (that I can think of) is happening. Either one of the parameters is in error (copied from the data sheet of a different model?), the manufacturer is using a non-standard definition of bandwidth or rise-time, or your scope has a severe under-sampling problem, and they are listing the bandwidth as the nyquist frequency which is below the 3 dB point of the AFE.
If you post the model of the scope, we can provide more useful information.