Belfast traffic-island death: the evidence behind a six-year sentence for a drink-driver

Paul Marshall was standing on a traffic island in Belfast city centre when a van mounted the refuge and struck him. The driver had made a lane-changing manoeuvre at speed and was subsequently found to have been over the drink-driving limit. Unlike many recent fatal-crash reports, where the police can only ask for witnesses, this case now has a criminal court outcome and a published account of the driving that caused the death.

On 3 July 2026, John Taggart, 36, from Belfast, was sentenced at Laganside Crown Court after pleading guilty to causing death by dangerous driving and driving with excess alcohol. He received a six-year sentence, with three years to be served in custody and three on licence, and was disqualified from driving for seven years. The collision had happened almost two years earlier, on 2 September 2024. These are dated court facts, not speculation about an unresolved investigation.

What happened on 2 September 2024

The police account places the collision at around 5.40pm on a Monday in Belfast city centre. Taggart was driving a white Transit van. Paul Marshall, 70, was standing on a traffic island. According to the PSNI Collision Investigation Unit, Taggart moved left from the outside lane into the inside lane at speed, then mounted the island and struck Mr Marshall.

That sequence is central to understanding the case. Mr Marshall was not described by police as stepping into the carriageway directly in front of a vehicle or attempting to cross between moving cars. The confirmed police description is that he was standing on a refuge. The van came onto that refuge during the driver's manoeuvre. A refuge is a space designed to separate a waiting pedestrian from the normal path of vehicles; the protection it provides depends on vehicles remaining within the carriageway.

The PSNI said Mr Marshall suffered catastrophic injuries and later died in hospital. A reader should not have to encounter invented details about his final moments to understand the gravity of the event. The collision was fatal, his age was confirmed, and the court has dealt with the driver. That is enough to support serious scrutiny of the actions established by investigators.

The alcohol evidence, precisely stated

Police said Taggart failed a preliminary breath test after being arrested at the scene. In custody, he later gave a lower evidential breath reading of 95 micrograms of alcohol per 100 millilitres of breath, which police characterised as almost three times the legal limit. The word “lower” refers to the relationship between the custody result and an earlier reading; the published statement does not provide the exact preliminary figure.

Careless reporting can create confusion here by writing “95mg of alcohol” without the breath-test units, or by implying a blood-alcohol test was performed. The PSNI described a reading per 100ml of breath. The evidence also supports the fact of an excess-alcohol conviction, because Taggart pleaded guilty to that charge. It does not give licence to invent what he had drunk, where he had been, how long he had been drinking or what he said to officers.

The lane manoeuvre and the intoxication finding are distinct aspects of the case. Investigators described the vehicle movement that resulted in the impact. The court dealt with dangerous driving causing death and excess alcohol. Both belong in the report; neither should be collapsed into an unsourced claim about every decision the driver made in the period before the crash.

Why the traffic island is the most revealing detail

When the public hears that a pedestrian died in an urban collision, the implicit mental picture is often a person crossing in traffic. The documented position of Mr Marshall breaks that assumption. He was on an island, outside the normal through-running lane. The case therefore illustrates the limits of relying on a designated waiting space as protection against a vehicle that leaves its expected path.

But the same detail must not be misused to allege a defective refuge. The available official statement does not identify a road-design fault, deficient barrier, sightline problem or maintenance defect. There is no independent engineering audit in the material reviewed for this article. The confirmed causal conduct belongs to the convicted driver; any separate infrastructure lesson would require an evidenced assessment of the actual site, its design and the feasibility of additional protection.

It is reasonable, after a serious urban pedestrian death, to ask whether other refuges in the area have been reviewed against vehicle encroachment risk. It would not be reasonable to claim that a particular barrier would definitely have saved Mr Marshall, or that the road authority should have anticipated the specific manoeuvre, without detailed evidence of vehicle dynamics and the site. In-depth journalism can ask the hard question while refusing to invent its answer.

What the sentence means — and what it does not

Six years is the overall sentence announced by police. The published breakdown is three years in custody and three years on licence. Licence time is part of the sentence and comes with supervision and conditions; it is not an additional period of imprisonment. The seven-year disqualification concerns driving entitlement. It is separate from the prison-and-licence arrangement and should not be confused with the length of the custodial sentence.

A guilty plea means the case did not require a contested trial to establish those admitted offences. It does not mean a reporter can fill in unreported testimony. The PSNI summary sets out the offences, sentence and what investigators determined about the collision. Without the judge's full sentencing remarks, a journalist should not assert which aggravating factors produced precisely six years, describe the defendant's subjective remorse, or claim the sentence was increased or reduced by a specific amount.

These distinctions are particularly important when discussing serious crashes in public. A sentence often attracts the loudest reactions because the number of years seems short against the permanence of a death. The legal structure explains that time in custody and on licence are different, but it does not decide how a bereaved family experiences the outcome. Sensational commentary is no substitute for accurate legal reporting.

Where this case fits in the wider safety picture

Alcohol-related road harm has not disappeared in Northern Ireland. The 2025 Road Safety Strategy monitoring report recorded nine people killed in road collisions where alcohol or drugs were attributed as a causation factor, compared with 16 in 2024. That is a combined category and must not be labelled “nine drink-driving deaths”. The 2024 collision that killed Mr Marshall is not part of the 2025 annual figure.

Those figures also need context. An annual decrease does not mean a convicted drunk-driving fatality has become less important. Conversely, one serious prosecution cannot establish that drink driving is increasing. Road safety policy should compare long-term patterns with clear definitions, distinguish alcohol from drugs where data permit, and examine enforcement alongside collision outcomes. An individual court judgment provides certainty about one offender's acts; official statistics provide broader context, not an explanation of the individual offence.

Accountability after a completed prosecution

This is a case in which a central accountability question has already been answered: the driver admitted dangerous driving causing death and driving with excess alcohol, and the court imposed a custodial sentence and driving ban. The remaining public-interest questions are different from those in an active collision inquiry. What can be learned about protecting pedestrians who are stationary on designated refuges? How are high-consequence urban road departures assessed? How can the evidence from convictions strengthen prevention without falsely implying that every refuge is unsafe?

The published PSNI account does not establish that such a review took place or that a site-specific engineering remedy was proposed. This report does not claim to have commissioned one. It demonstrates why careful reconstruction of confirmed facts should come before policy slogans: a particular driver, a documented lane movement, an incursion onto an island, a 70-year-old man killed, and a court outcome.

The verified story is neither an unsolved mystery nor merely a statistic. It is an established fatal driving offence with a clear sequence of events. Paul Marshall was killed while standing on a traffic refuge; John Taggart pleaded guilty; and on 3 July 2026 a court sentenced him to six years and banned him from driving for seven years. Reporting those facts accurately, while identifying the unanswered engineering questions as questions, is the standard this case demands.

Sources: PSNI sentencing account of 3 July 2026; PSNI road-safety statistics; 2025 Northern Ireland Road Safety Strategy monitoring report; PSNI collision advice.