To mark the 50th anniversary of NWP in the UK, the Met Office hosted the Royal Meteorological Society's national meeting on 18 November 2015, chaired by Ken Mylne (Met Office). The Met Office's Chief Executive Rob Varley opened the meeting, saying that synoptic-scale NWP in mid-latitudes constitutes one of the crowning scientific achievements of the twentieth century. The Met Office's Chief Scientist Julia Slingo then expressed optimism about the future, and paid tribute to Malcolm Walker, Chair of the Royal Meteorological Society's History Group, who was unable to attend the meeting. The main talks covered past, present and future, with recurring themes of progress being driven by improving science and technology, significant weather and world events, the importance of translating science into services and of communicating forecasts of the weather and its impacts. Anders Persson (Uppsala University) described the pioneering men and women behind NWP in the UK between 1948 and 1965. He also contrasted the differing approaches of the UK, whose two-layer baroclinic model with associated vertical motion was needed to diagnose rainfall, with other centres in the USA and Sweden, who were initially content with simpler one-layer barotropic models. The baroclinic model's higher memory overheads led to a smaller area, which led to too rapid propagation of boundary errors, while the 2h timestep used in the numerical solver led to instability. On 2 November 1965 the new Director-General John Mason committed the Met Office to operational NWP at a press conference. Nick Grahame (Met Office) covered the next 50 years, describing the Octagon and Rectangle models, the UK's global model (operational in 1982, catalysed by the Falklands War) and ECMWF's first ensemble in 1992. He showed the first UK operational NWP forecast and included quotes from former forecasters that NWP was initially ‘a useful idea to consider’ or ‘mostly ignored’. By the late 1970s it had become an important ingredient. The Great Storm of 15/16 October 1987 motivated the creation of the National Severe Weather Warning Service and big changes in how NWP was used and forecasts communicated. The benefits of coordinated communication of weather warnings were particularly apparent in the Burns Day storm (25 January 1990). More recently, the St Jude's Day storm (28 October 2013) had its first Amber warning issued five days in advance and received much media coverage, including of the uncertainty. Peter Clark (Reading University) described convective-scale NWP. Today's forecast problems (inland shower penetration, fog etc.) involve small-scale weather and require non-hydrostatic models allowing for vertical motion when modelling convection. At kilometre scales, these models are convection-permitting, resulting in output that looks realistic but is usually wrong in detail. The Met Office's mesoscale model was developed from 1976 and was gradually introduced into operations from 1984. A non-hydrostatic dynamical core was then introduced in the Met Office Unified Model (MetUM) in 2002. The Boscastle flood on 16 August 2004 showed that the kilometre-scale configuration provided a much better representation of the rainfall than the operational 12km configuration. The MetUM was subsequently upgraded to a 4km grid in 2005 and then to the present 1.5km grid in 2010. A 300m grid was used for the Weymouth Bay sailing events in the 2012 Olympic Games, and a similar configuration is now being trialled for the London area, including Heathrow Airport. Brian Golding (Met Office) spoke about applications of NWP. The first operational application was route-planning for aviation, which required jet-stream predictions. The extraction of North Sea oil and gas led to the launch of the first swell model in 1976, which was subsequently upgraded to an ocean wave model in 1979. Likewise, the Thames Barrier construction motivated the development of the storm surge model with the Institute of Oceanographic Sciences. The MetUM now supplies rainfall output to the Centre for Ecology and Hydrology's Grid-to-Grid model for Environment Agency flood prediction. The Chernobyl nuclear accident in 1986 led to the development of the NAME model, which is used to track the spread of nuclear and chemical material, volcanic ash and airborne diseases. Health applications were also presented along with hazard forecasting, which combines the meteorological risk with the impact of an adverse outcome. The WMO's High Impact Weather initiative aims to look at all aspects of weather impacts to create a more resilient world, including evaluating what warnings to issue and whether people respond. Such risk modelling is inherently probabilistic, with ensemble forecasting at its heart. Dale Barker (Met Office) spoke about plans for the next 5 years and speculated about the next 50. Satellite data dominate other observations in their impact on forecast skill. Ensembles continue to grow in importance, but will need to be used in conjunction with deterministic models. New verification measures are being developed to better describe the added value of the UK model over the global model, which is increasingly getting better at forecasting weather parameters. Future upgrades will need to be tested over longer periods because convective-scale models are inherently variable in their skill. The Met Office is installing new supercomputer and storage infrastructure and is considering which combination of longer forecasts, larger ensembles, higher resolutions and increased complexity (e.g. coupling atmosphere and ocean) will best deliver the anticipated £2bn benefit to the UK. Improvements to data assimilation will continue, enabling hourly forecast updates. The big challenge is computer architectures, which cannot get faster at their current rate because the power consumption becomes prohibitive, so improved parallelism is the key to future performance, driving the redesign of NWP software at the Met Office. An animation of a fog forecast showing a remarkable consistency with satellite observations for 2 November 2015, exactly 50 years after Met Office operational NWP started, showed the level of progress that has been achieved. The meeting closed with a poster session showing recent developments and demonstrations of the Weather Observations Website (http://wow.metoffice.gov.uk) and Informatics Lab (http://www.informaticslab.co.uk).