How often are these flights late or cancelled?

EDI ↔ SOUTHAMPTON

In the twelve months from July 2025 to June 2026, 29.6% of 1,051 movements arriving into EDI were disrupted — that is, they landed more than fifteen minutes late or were cancelled at short notice.

Put the other way round, 70.4% of 1,051 movements arrived on time by the industry's fifteen-minute standard. Severe delays were rarer: about one arrival in every 263 crossed the three-hour mark. Lower is better on every figure on this page. These numbers describe what happened on the route; they are not a prediction for any individual flight.

The four numbers at a glance

In the twelve months from July 2025 to June 2026, 29.6% of 1,051 movements were disrupted and 1.3% of 1,051 movements were cancelled at short notice.

When a flight was late, it was typically late by the range shown below. The delay is given as a range, not a single number, because the regulator rounds its published percentages and never says how many flights landed exactly on time.

When it is late, how late is it?

Most delays here are short. The ladder below shows how many arrivals crossed each threshold, from a quarter of an hour up to six hours; each rung can only be smaller than the one above it.

Cancellations are counted separately: 1.3% of 1,051 movements. “One in N” means how many arrivals you would typically see between events at that level.

Slightly late, or occasionally a disaster?

When this route runs late, it is usually only a little late: big blow-ups are rarer here than on the typical UK route.

The histogram below counts every recorded arrival, from more than fifteen minutes early through to more than six hours late. A tall bar near the middle and a thin right-hand tail is the healthy shape.

Has this route got better or worse since 2018?

Best month: Jan 2020, 4.2% of 96 movements. Worst month: Jul 2022, 63.2% of 106 movements.

The line traces every month in the archive; lower is better. Gaps are months with no published data, never zeros. The 2020 and 2021 readings reflect the pandemic, not normal service.

What did the big disruptions actually do here?

Each named event below is shown against this route’s own baseline for the twelve months before it, and against how long the route took to get back to that baseline.

A month worse than baseline is not proof the named event caused it: the open data carry no delay-cause field at all. Some events sit near the edge of the archive, which limits how long a recovery can be traced.

Which time of year has historically been hardest?

July has historically been the hardest month on this route: in a typical July, 37.6% of arrivals were disrupted, measured across 6 Julys in the archive. February has been the calmest, at 15.3% across 7 Februarys.

Even July has ranged from 29.3% to 63.2% in different years, so this is a record of what happened, not a forecast of what will. Lower is better throughout.

Each cell is one calendar month of one year; the number is the share of that month’s arrivals that were disrupted. Darker means worse. A blank cell means no published data, not zero.

Which airline has the better record here?

Only one carrier published enough flights here to report: Loganair Ltd, 29.6% of 1,051 movements disrupted.

Figures cover arrivals, July 2025 to June 2026; lower is better in every column. Carriers with fewer than twenty published flights are suppressed rather than shown as zero. The page makes no ranking: the operations behind these shares are not matched for aircraft, time of day or season.

The noise check is a descriptive approximation assuming independent operations. Delay correlation and mix differences are not estimated; no ranking is derived.

Would another UK airport have been better?

Among UK airports flying to this city with enough flights to publish, NCL had the lowest disruption share at 12.9% of 629 movements and ABZ the highest at 46.2% of 39 movements — small sample, not ranked. This route sits at 29.6% of 1,051 movements.

Each figure is arrivals into that UK airport, all carriers combined, over the same twelve months. It is a historical record, not a recommendation about where to fly from, and it says nothing about getting to or from each airport on the ground.

Does the direction you are flying change anything?

Barely: arrivals into EDI ran 29.6% of 1,051 movements disrupted, departures 28.6% of 1,052 movements.

It matters legally as well as practically: a departure from the UK and an arrival into the UK on a UK or EU carrier fall under different conditions, and a departure delay does not establish a late arrival at your final destination.

How many flights here crossed the three-hour mark?

The table below counts arrivals and departures that ran more than three hours late — the threshold that is one of the conditions for compensation — across the whole archive and within the last six years.

The pound figure beside each count is arithmetic, not money anyone received: it is the count multiplied by the maximum statutory amount for this distance band. Whether any of those flights actually qualified is not knowable from this data.

The facts about this route

origin_destination is the final point of the CAA service, not evidence of a nonstop connection. The distance joins the coordinates of the named airports; the actual track is unknown. Frequency describes published operations.

What this data cannot tell you

Delay cause is not published in the regulator's open data. Monthly aggregates describe a population, not a specific flight. Cells below twenty operations are suppressed; twenty to forty-nine are descriptive only and not ranked. Arrivals and departures do not add up. Cancellations are short-notice only, by the CAA definition. A movement is not the same as a scheduled flight.

0 operations (0.00%) excluded because their published percentages could not be reconciled to whole numbers. Excluded operations enter neither the numerator nor the denominator. Outcome figures and comparisons are unavailable when the excluded share exceeds fifteen percent.

Quick answers

How this was calculated

Your trip has already gone wrong — what now?

Related records