Minuit errors and Up value


With minuit, I am trying to minimize a multi-parameter function (4 free parameters and 65 data points). The supplied function is a chi2 function.
However, I am a bit confused on how to estimate the correct errors with Minos after reading the last part of the pdf “Interpretation of errors with Minuit”. More specifically with the Up table as a function of Confidence level and Number of parameters (Table 1.3.3).
So here is the question: should the Up value provided with the user function be a function of the number of free parameters (4) or the number of degrees of freedom (65-4-1) ?

Thanks for any input.