Recent advances in Natural Language Processing (NLP) and Large Language Models (LLMs) show significant potential for automating the extraction of process information and the generation of process models directly from descriptions. We present an approach that automatically generates Business Process Model and Notation (BPMN) from process descriptions using LLMs and Prompt Engineering. Our approach considers a large subset of BPMN elements for generation and focuses on the quality of the generated models that are evaluated with Graph Edit Distance. We propose a prompting strategy that serves as a guiding method to help researchers in the future use of LLMs in the field and provide a tool to support the approach. The results evaluated through a quantitative comparison show that our approach outperforms a state-of-the-art tool, especially for more complex processes.