All regulation imposes costs, both monetary and nonmonetary, on a range of agents, as Whitney and Schneider [1] rightly point out. For instance, the research ethics systems as they are currently operating in most industrialized countries impose costs on researchers, research sponsors, future patients and the state. In assessing the costs of the research ethics regulation system and deciding whether they are justified, it is important to distinguish between two different aspects of the system. We need to distinguish between first the regulations themselves and secondly the system for ensuring that the regulations are followed. The importance of this distinction can be highlighted by the following example. Consider a multicentre, randomized, controlled, phase III clinical trial of a first-in-class new chemical entity. The trial involves a number of outpatient visits, collection of blood samples and two computed tomography scans that are additional to the normal clinical follow-up for the condition in question. Because of the complexities in negotiating the research ethics system, the time from submission of the protocol to final approval is 100 days. After the start of the trial, recruitment proceeds slowly because many potential participants do not consent to participate and a sufficient number of participants are only recruited 100 days after the originally estimated time for end of recruitment. In this example, we have two 100-day delays. One delay caused by the research ethics system and one caused by (specific provisions within) the research ethics regulations. Should we conceive of these two delays as equally problematic? Clearly not! If the necessary level of regulatory scrutiny prior to approval of the protocol can be achieved by a simpler, more efficient or faster research ethics system, then there is no reason not to implement the simpler system and reduce the regulatory delay to a minimum. However, the fact that potential participants will not consent does not even provide a prima facie reason to consider abolishing or relaxing the requirement for consent. The requirement for consent is justified in this case by a number of overlapping ethical and legal principles and would exist as a moral and legal requirement even if we had not implemented any research ethics system to enforce it. Most legal systems have rules against bodily invasion without consent. In the United Kingdom, for example, this would amount to the tort of battery [2]. The research in this example imposes risks, burdens, bodily intrusion and cost on the participants, and some of these would have constituted actionable breaches of their established human rights if they had not consented. In a European context, this would follow both from the 1950 ‘European Convention on Human Rights and Fundamental Freedoms’ [3] and from the much more detailed 1997 ‘Convention for the Protection of Human Rights and Dignity of the Human Being with regard to the Application of Biology and Medicine: Convention on Human Rights and Biomedicine’ [4]. Thus, performing this research without consent is not a trivial matter. Furthermore, if the research subjects had not consented and had not been told that they had participated in research, they would have been deceived. Researchers who performed this type of research without consent before the establishment of research ethics systems may not be considered blameworthy because they acted in ways that were customary at that time, but they were still morally wrong in not seeing that consent was required. However, if the cause of the recruitment delay can be traced not to consent itself but to specific requirements concerning the form or process of consent, we may of course have reason to consider whether these specific requirements can be justified. Similarly, we may consider whether the same consent requirements are justified for different kinds of research. However, that there are clinical research projects that require consent is indisputable. As for most (or perhaps all) moral requirements, the consent requirement can be defeated by more important moral considerations. The fact that a requirement is defeasible does not make it nonexistent. For instance, the point of conscription is that we know that involving people in certain activities normally requires their consent, assent or at least acquiescence, but we think we have overwhelming reason to deviate from that requirement in specific circumstances. Returning to the question of the justifiability of the research ethics system, what are the goals that the system is supposed to achieve and that, if achieved, can justify the system? A first and perhaps naive answer is that the goal of the research ethics system is to prevent unethical research from taking place. This clearly requires further clarification of the term ‘unethical research’ that is rather vague. There are numerous ways in which a particular research project can be ethically problematic, but in the present context, I will focus on those cases in which the ethical issue concerns the way research participants are treated by the investigators. Therefore, ethical considerations about the goals of the research, its methodological quality, or the researchers’ conflicts of interest will be excluded from the present discussion. With this qualification, the four main ethical issues that arise are in relation to research in which: (i) research participants are deceived; (ii) research participants are exposed to additional risk; (iii) research participants are exposed to burdens, including bodily intrusion; or (iv) the rights of research participants are breached. Thus, the function of a research ethics system cannot be confined to protecting research participants from physical or psychological harm. Participants may not be harmed, or harmed significantly every time their rights are breached, but they may have been wronged. It is noteworthy that protection from physical harm is a task which, in relation to clinical trials, is often performed not by research ethics committees but by medicines regulatory agencies in many countries. These ethical issues that have been identified can be managed in different ways; for example, through fully informed consent as a device by which individuals voluntarily assume risks and burdens, permit bodily intrusion and allow researchers to act in ways that would otherwise breach their rights. Why do we need a system to ensure that this happens? Why can’t we simply put forward a set of reasonable ethical principles and specific rules and allow researchers to decide how to implement them in their particular research projects?1 The main reason is that the interests of researchers and research participants are not identical in the research context and that we do not require the researchers to act as fiduciary agents for the participants (i.e. only to act in ways that promotes the participants’ interests)2 [5]. This does not deny that researchers and research participants have identical interests in their role as citizens, or in relation to their potential to become future patients. However, the interests a person has in this role does not exhaust his/her total set of interests. I am a citizen, but I am also a husband, a father, a person with particular life goals, degree of risk aversivity and, possibly idiosyncratic, fears and apprehensions. In the research context, at least some researchers have interests related to career advancement, fame, recruiting premiums or making the research process run as smoothly and quickly as possible [6]. However, these interests may not be shared by research participants who may have interests in avoiding risk and inconvenience. If we allowed researchers to interpret research ethics principles by themselves, it would be surprising if they did not, at least sometimes, give slightly too much weight to their own interests and too little to the interests of participants. In addition, this would be even more likely to occur if we allowed them to redescribe their personal interests as pure scientific interests related to the acquisition of knowledge. If we have learnt anything from recent studies of the effects of conflict of interest, it is that even the best educated and most upright researchers are not immune to the pernicious influence of such interests. The design, conduct, analysis and reporting of research have all been shown to be susceptible to subtle subversion by interests competing with the researchers’ (and citizens’) interest in producing the most valid and unbiased knowledge [7]. Therefore, we need some way of protecting potential research participants from unscrupulous and self-interested researchers. This does not in itself entail that we need anything like research ethics committees. There are many other possible regulatory options, and indeed too many to review here. There are two options that seem attractive, because they would allow us to abolish research ethics committees. First, self-regulation has been suggested as a possible option; we should let the research community deal with researchers who cannot distinguish between their own interests and the interests of research participants. However, this is unlikely to work because self-regulation without enforcement rarely works, and because we have good historical evidence that self-regulation specifically in research is not effective [8, 9]. Self-regulation has not abolished scientific fraud and did not eliminate unethical research before the advent of research ethics committees. A second suggestion is that we should enforce research ethics regulations retrospectively through the courts. We should create a system in which a research participant can take an investigator to court for breaches of research ethics. However, this is unlikely to work unless our research ethics rules are very specific and not open to selective implementation by researchers. Furthermore, it is unlikely to be effective unless research participants are helped with the process of making their claims.3 We know that in the area of clinical negligence, many valid claims are never put forward because of the perceived complexities and costs of the legal process [10]. Researchers might also object to such a system because they could not proceed with their research in a context of legal certainty that it was permissible. But what about the research ethics rules themselves, are they too strict? It would be foolish to claim that the current set of regulations applying in any particular jurisdiction are perfect, and there are reasons to believe both that some rules are too strict4 and that some rules are applied too strictly [11, 12]. Again in the context of clinical research and consent, there is no inherent reason why all participant information sheets and consent forms have to be written in the same style, have the same lay out or use the same standard phrases. For example, the proposition that research participants can withdraw from research without negative consequences for themselves can be given linguistic expression in many ways that are extensionally equivalent and there is no justification in principle for a research ethics committee to insist on one, and only one, ‘correct’ way of expressing this. However, it is easy to see how committees end up making such unjustifiable demands, driven by a belief that all decisions need to be consistent, and wrongly believing that consistency is, in this case, a matter of linguistic conformity. Maybe all our research ethics regulations should therefore be prefaced with the following statement: ‘The main task of the research ethics committee is to ensure that research is conducted in an ethically appropriate manner, taking into account the specifics of each individual research project. Strict conformity to the letter of the rules may not always be the best way to achieve this.’ The author has no conflicts of interest to declare.