The size of the membership set is investigated in the presence of bounded disturbance and l₂ bounded parameter uncertainty. A tight upper bound of the diameter of the membership set is derived in a deterministic setting, which indicates that the diameter does not converge to zero in general. Then, a probabilistic upper bound of the diameter is derived in a stochastic setting, where the disturbance and the parameter uncertainty are assumed to be random variables and to take a value near the worst ones with a probability. This probabilistic bound leads to the fact that the diameter converges to zero with probability one as the number of samples tends to infinity.