The first recorded archaeological observation about Croft was that by John Aubrey in his Monumenta Britannica (1674?) where he noted that a large double-ditched camp, 'the Ambry', was to be found within 'Crofts-parke'. The first archaeological fieldwork in recent years was also at Croft Ambrey hillfort. This was the campaign of investigation by Stan Stanford of Birmingham University Extra-Mural Department between 1960 and 1966 (published as the monograph Croft Ambrey in 1974).

The aim in developing an archaeological programme for the Croft Estate was, from the outset in 2000, to conduct a full survey of the estate lands, and to devise a programme of closer investigation of the house environs and of the prehistoric landscape surrounding Croft Ambrey. Two years on, the survey is complete, and the exploration of the close environs of the mansion is at an advanced stage. The prehistoric project, meanwhile, is still at the planning stage.

The prehistoric farmsteads and field systems that survive here have now been extensively documented, and understanding of the complexity of Croft Ambrey itself has increased dramatically. For instance, a huge area of north-facing hillside within the defences of the fort appears to have been deliberately scarped by the Iron Age community to create stances and terraces for structures. What had been described as an 'annexe' to the fort was found to be a primary element of the enclosure sequence, with a length of west-facing bank and ditch cut through by the later massive defences. The inner entrance above the south-western outer gateway excavated by Stanford seems to have been made much more elaborate in its later phases, to make this one of the most heavily fortified entrances to a hillfort known anywhere in Britain.

One of the most interesting findings in Stanford's excavations was a complex and long-lived shrine, mainly of Romano-British date, sited on the slope outside the highest inner rampart. The possibility has emerged also that a series of earthworks on the inner scarp of the interior quarry ditch of this same rampart are of Romano-British date. As such they may represent the presence of further shrines, indicating that after abandonment, the site became an important focus for continuing religious activity.

New detail can also be added to what is known of the late medieval or early post-medieval use of the hillfort within Croft Castle's deer-park. While the pillow-mounds at the site were already well known, the recent survey has revealed what appears to be a warrener's house within its own enclosure below the fort, and traces of saw-pits on one of the outer banks.

Meanwhile, other unusual features have been found elsewhere during the survey, that throw light on activities that have yet to be documented well within the county. An example from Bircher Common is provided by small earthen platforms sited close to streams. These platforms may mark the sites of medieval transhumant farming shelters. Besides these traces, up to a dozen sets of earthworks belonging to small homestead-type complexes of the post-medieval period have also been located on the Common. The traces are the remains of houses and barns within small enclosures, in one case with their lazy-bed cultivation plots surviving fossilised within an outer enclosure. Elsewhere on the Common are the extant earthwork remains of no fewer than four sub-rectangular farmstead enclosures of presumed late prehistoric or Romano-British date, with traces of their contemporary fields.

The survey therefore provides a datum for future study, but more particular concerns in reference to management are the establishment of evidence for former planting patterns, and the possible restoration of areas of former wood pasture.

In September 2001, then, three weeks of investigation and fifteen small trenches revealed the mode of construction (and later demolition) of the formal gardens that once provided the immediate setting of the mansion. The sequence was elucidated in large part thanks to advice during a site inspection visit of the excavations from a group of specialists. These included Paul Stamper of English Heritage, David Whitehead of the Hereford and Worcester Gardens Trust, Brian Dix of Northamptonshire Archaeology (the excavator of Kirby Hall and of The Privy Garden at Hampton Court Palace), David Jacques of De montfort University, Leicester,and the Trust's own Historic Gardens Adviser, Katie Fretwell.

The stone foundation of the semi-circular stair was found in the excavations, and finely-dressed pieces of the local limestone that once formed the curving steps themselves were found in the rockery within the nineteenth-century walled garden to the north-west of the castle. Beyond the upper sloping garden to the west were found wide grooves cut into the bedrock, perhaps to provide soil depth for the bedding in of soft fruit trees such as apricots. Most of these features had been evident from aerial photographs taken during drought conditions in the summer of 1995.

The lower sloping garden also terminated in a level terrace, but this has retained its original form, having been constructed entirely as an earthwork without revetment walling. A divided grassed ramp leading down to the dam between two formal pools and into the southern formal broadwalk is thought most likely to date between 1700 and 1710. It therefore seems most likely that this southern sloping garden, and a parallel east garden continuously sloping southwards from Croft Church, together with an attached curving prospect walk, were added during a major expansion of the garden in these years.

Beneath all these features were found traces of medieval activity. This included fish-ponds in a side-valley that had previously been thought to have contained a formal cascade to the west of the garden. There were also finds of fragments of medieval domestic wares, and pits containing stone demolition deposits (but no brick and mortar or other post-medieval material) sealed at depth beneath the later garden features.

West of the castle, further traces of the gardens of 1700-10 were found, in the form of the stone-built basal courses of garden walls, a foundation for the plinth for a statue, and an artificial earthen terrace. Some localised subsidence had been made good by the dumping of deposits of late seventeenth century brick and window glass. The deposit was dated by the inclusion of a near-complete stamped clay pipe made locally at Pipe Aston between the years 1690 and 1710. The window glass included hundreds of fragments of small window panes made from opaque glass. However, it also included numerous large pieces from medieval stained glass windows, presumably re-used in the later casements.

Below this dump deposit was found a sterile yellow clay deposit. When this had been removed, a marked depression was noted running obliquely across the excavated area. This was found to follow the line of a robbed-out wall foundation. In the back-fill of this robber trench there were numerous pieces of carefully dressed stone from a large medieval arch, along with other moulded pieces of medieval stonework. This was an important find, since it appeared to bear out the theory that we had developed, that the medieval Croft Castle perhaps once stood to the west or north-west of the present castle, at a greater distance than at present, from the parish church of St. Michael. Meanwhile, across the whole of the excavated area dumps of late sixteenth century brick and tile were found. A deposit of mortar below these dumps nearby was thought initially to represent a medieval floor level.

It appears that this later wall therefore represents a conversion of the medieval castle into a fine if small-scale late Elizabethan mansion, presumably in Sir James Croft's time. Sir James had been Comptroller of the Queen's Household, and died in 1590. This is therefore the building (a medieval fortified manor/small castle, converted into a modest Elizabethan mansion) that we can now assume was badly damaged in the Civil War, after Sir William Croft became the only Head of a leading Herefordshire family to have been killed in battle. Traces of burned timbers, however, represent the demolition of this compromised structure, and the mortar floor was revealed to be a mortar-mixing surface when the ruins of the damaged house were used as a workshop area for the building of the new mansion in 1662-3.