Spray penetration and mixture formation in GDI engines are crucial to a reliable ignition and the subsequent combustion. For the prediction of the mixture formation process, computational fluid dynamic simulations are applied. Therefore, details about the nozzle exit conditions are essential, either as boundary conditions to be set, or to validate the numerical results.This paper presents experimental results on the influence of boundary conditions on the spray structure at the nozzle exit of a GDI injector. The injector investigated is an outwardly opening piezo injector, generating a hollow cone spray with a string structure. The distribution of the strings (the so-called "string structure") is needed for the starting conditions of the computational fluid dynamic simulations, as the origin of the strings is unresolved so far. The experiments are conducted in a constant volume pressure chamber varying the ambient temperature between 300 and 600 K and the ambient pressure between 0.1 and 1.0 MPa. Analyzing the impact of the different injection conditions, the injection pressure is varied between 8 and 12 MPa, utilizing energizing durations between 200 and 600 μs. In addition, the needle lifts are varied, too. Visualization techniques are used to investigate the spray structure and the penetration length.The evaluation of the results is divided into two parts. First, spray images of the nozzle exit flow are compared concentrating on general parameters of the spray such as the penetration length. In the second part, the string structure is investigated by extracting the light intensity profile at certain nozzle distances. The analysis of the spray structure shows that the ambient and injection pressure as well as the ambient temperature influence the circumferential fuel distribution at the nozzle exit; the needle lift shows no significant influence in the visualization results. Comparing the influence of the different parameters on the string distribution shows that the injection pressure and the ambient temperature have the greatest influence. But even though the greatest influence can be observed varying the injection pressure and the ambient temperature, these influences are very small.