Current design codes like the Eurocode use safety or reliability classes to assign target reliabilities to different types of structures or structural members according to the potential consequences of failure. That, in essence, is a risk-based criterion. A wide range of structures is designed with such codes, and distinction is made between reliability classes. These reliability classes are not necessarily well suited for flood defense systems, neither are the design rules and partial safety factors, which are calibrated for a wide range of standard applications. For a flood defense system protecting a large area from flooding, on the other hand, it is worthwhile to base the design and safety assessment standards on a risk assessment - a tailor-made solution. The investments can be considerable and the stakes are high, especially for low-lying delta areas, where the consequences of flooding can be devastating. In order to answer the question “How safe is safe enough?” a framework for acceptable risk is required. Subsequently, from acceptable risk we can deduce target reliabilities for the protection system as a whole as well as for its elements. For practical application, these target reliabilities can then be translated into design and assessment rules; for example, using LRFD (load and resistance factor design) to derive partial safety factors. This paper describes how to define safety standards for flood defenses, in particular dikes, step-by step. An important aspect in translating high-level requirements into specific (low-level) design rules that apply to specific failure modes for specific flood protection elements is the so-called “length effect”. This is especially relevant for long-linear structures like dikes, where usually the length is much larger than the scale of fluctuation of dominant load or resistance properties. The longer the structure, the higher the chance to encounter either and extreme load or a weak spot (i.e., low resistance) – hence the word “length-effect”. The effect is that the probability of failure increases with the length of the dike. The implication for design and assessment rules is that the reliability requirements to a cross section (“zero length”) need to be stricter (i.e., higher target reliability) than for the whole reach. This paper attempts to demonstrate how tailor-made safety standards for large scale flood defense systems can be derived in a risk-based fashion. Since flood defenses differ from smaller scale geotechnical structures in many aspects and given the volume of investments in such large-scale engineering systems, it is very attractive to deviate from the standard design codes. That is not deviating conceptually, but rather deriving safety factors for the specific application to better account for the characteristics and uncertainties involved. The authors strive to show that safety levels and partial safety factors in the presented approach are far from arbitrary. They are part of an overall consistent flood risk framework, a framework that provides a link between geotechnical engineers and other disciplines involved in providing safety from flooding.