INTRODUCTION Safety standards for consumer products have traditionally been structured to apply to classes or groups of goods such as prams, toys and cots. These are known as vertical standards. Recently there has been interest in developing standards which are hazard based and cover broader classes of goods. Such standards are referred to as horizontal standards. The Working Draft European Horizontal Standard for Children's Use and Care Articles takes a horizontal approach to a class of products. This paper considers how horizontal and vertical approaches can be mixed, with specific reference to toys, but also considers the broader question of how this sort of decision can be made more generally. OBJECTIVES• To consider whether the coverage of the Working Draft European Horizontal Standard for Children's Use and Care Articles (CENrrC 252 WG6 N 145) could be expanded to cover toys. •To develop an approach for making this judgement based on injury surveillance data. • To identify issues which need to be addressed in deciding the appropriate mix of horizontal and vertical standards. METHOD A random sample (n = 1060) of toy-related injury cases among children aged 0 to 9 years presenting to sentinel emergency departments in Australia was examined. Cases were drawn from the National Injury Surveillance Unit's Injury Surveillance Information System (ISIS). Cases were drawn at random from all records that had been assigned anyone of more than 40 toy codes. Bicycles were excluded due to the difficulty of separating out small toy bicycles from larger bicycles used for transport. The description of the injury circumstances for each selected case was carefully examined. Cases were only included for further study if the description indicated that some particular feature of the toy contributed to the injury. For each of these cases, injury events were classified according to the major categories of hazard covered by the Working Draft European Horizontal Standard for Children's Use and Care Articles. In addition, two extra hazard categories were added to cover instability and Joss of control of wheeled toys, and projectile toys. The types of events occurring in each of the major hazard categories were closely examined to assess whether the provisions in the draft horizontal standard might be suitable. RESULTS There is a considerable overlap between the provisions of the draft horizontal standard and those needed to deal with the causes of injuries related to toys. There are, however, aspects of toy-related injury, including mobility, energy producing characteristics, desired use patterns and size of the items, that pose different hazard management questions. Hazard patterns change dramatically with age. Injuries often occur when young children access toys designed for older children. Major hazards are associated with height, shearing and cutting, wheeled toy instability and loss of control, ingestion, inhalation and insertion, and the strength and durability of toys. CONCLUSIONS A key issue which has emerged is the need to adequately define the boundaries of applicability for a horizontal standard. Examination of empirical evidence of hazard patterns can provide guidance and is a necessary step in boundary definition. After undertaking this analysis of injuries associated with toys it is concluded that: • the philosophy and many provisions of the child use and care standard are applicable to toys; •toys differ from child use and care items in that necessary characteristics create hazards for younger children; • the management of toy-related injury may place more emphasis on information than is envisaged for child use and care items; the information task for toys is more complex in that it requires communication to users of the risks to younger children who are not intended to be users; • a vertical standard for toys is necessary because of these differences.