This paper investigates the event-triggered distributed model-free iterative learning bipartite consensus (ETDMFILBC) problem for unknown nonlinear discrete-time multi-agent systems. A dynamics linearization data model is first established by employing a compact form dynamics linearization approach, and the estimation algorithm of its parameters is also formulated. After that, the output observer and dead-zone operator are designed. Meanwhile, applying the signed graph theory, an ETDMFILBC scheme is proposed. Moreover, the convergence property of the proposed algorithm is strictly proved using the constructed Lyapunov function. Finally, the results of the simulation further verify the correctness and effectiveness of the designed scheme.