Editing the symposium contributions has been an in-teresting and instructive experience, because they providea broad overview of the well-nigh unlimited creativity ofmarine scientists to try and find building blocks for anecosystem approach to fishery management. A wealth ofideas appears to be emerging on how different aspects of thestate of exploited marine ecosystems can be measured.Although the metrics produced are commonly referred to as‘‘ecosystem indicators’’, however, it is generally less clearwhat they actually indicate.The concept that management must take the widerranging effects of fisheries on the ecosystem into account iffisheries are to be sustainable without unduly affecting themarine environment is an important step forward. Also,these metrics may help to define management objectivesmore clearly. However, the contributions have not con-vinced me yet that the quantitative indicators developed areof much pragmatic use in controlling the fisheries. Part ofthe problem lies in the lack of rigour used in the definitionof what an indicator is, and what purpose it is supposed toserve. New metrics of ecosystem attributes appear to beadded to the list of potentially useful indicators for fisheriesmanagement, without a clear idea of what they actuallyindicate. Metrics, however, measure something specific,while indicators are supposed to tell us something differentfrom what they actually measure. The tacit inference isoften made that the metrics proposed provide an indicatorof the rate of overexploitation, but little if any evidence isprovided that they do. In fact, defining ecosystemexploitation rate is a problem in itself, because how maydifferent fisheries be combined in a single ‘‘pressureindicator’’ at an ecosystem level? Can we simply addcatches or average fishing mortalities of pelagic anddemersal species, or must weighting factors be applied?Can we use pressure indicators from modelling ap-proaches? The empirical correlation between ecosystemindicators and empirical pressure indicators has beenlargely neglected, and if it has been addressed, the resultshave not been unequivocal. Still, this is crucial if thecriteria defined for selecting a suite of appropriateindicators are to be applied rigorously.A comparison with traditional, single-species manage-ment may help to clarify the issue. The chance that a fish iscaught (fishing mortality) reflects the impact of the totalfishery on the stock and is on theoretical grounds a pressureindicator (of fishing effort). Changes in relative stockbiomass represent a measure of the ‘‘well-being’’ ofa population, encompassing all anthropogenic impacts(through fishing mortality) and natural sources of variation(through growth and recruitment, including competitionand predation effects). Stock assessment is aimed atelucidating the interactions between these two indicatorsso as to formulate best possible advice. This is important,because without a pressure indicator no-one could givequantitative advice on the extent to which effort should bereduced, and without a well-being indicator, we would notknow whether fishing would do any harm.I tend to believe that the indicator approach to take intoaccountecosystem effectsinfisheriesmanagementwillonlyhelp if we are able to go beyond the statement thatecosystems are overexploited and unhealthy, and insteadprovideadviceonquantitativemeasuresforspecificfisheriesto resolve a well-documented problem. In fact, many of theproblems encountered in an ecosystem context may besolved pragmatically from a single-species, community, orsingle-fishery perspective (as currently pursued in theAlaska region). For instance, bycatch rates of marinemammals, turtles, and elasmobranchs are effectively a sin-gle-species problem that may be resolved along traditionallines of single-species assessment and management, buttaking into account that specific fisheries are often causingthe problem. To control impact on benthic habitats, we donot necessarily require a full ecosystem approach, but againa fleet-orientated approach is essential. The remainingproblems that refer truly to the ecosystem level, but havetoberesolvedbymanagementandwhichcannotberesolvedby traditional management approaches, relate to the balancebetween the fisheries exploiting different components.Before metrics of ecosystem structure and functioning maybeappliedasecosystemindicators,however,wefirsthavetoestablish a clear relationship with suitable pressure indica-torsofecosystemexploitation.Thismaynotbeeasybecausegood ecosystem metrics should integrate all sources ofvariation over a range of time delays, and therefore areunlikely to meet important criteria for good indicators likeresponsiveness, sensitivity, and specificity (sensu Rice andRochet, 2005). Meeting these criteria probably would bereserved for good pressure indicators, but much more workappearstoberequiredtodefinethemandtounderstandtheirrelationship with indicators of ecosystem health. The