A method is described for calculating verified error bounds for zeros of a system of nonlinear equations f(x) = 0 with continuous f : ℝ → ℝ . We do not require existence of partial derivatives of f, and the function f may even be known only up to some finite precision ε > 0. An inclusion may contain infinitely many zeros of f. An example of a continuous but nowhere differentiable function is given. n n