Delhi prison security is built around layers of personnel, surveillance and physical controls, but the escape history of Tihar shows why security in a large prison system is not a single barrier that can simply be strengthened. It is an institutional system in which staffing, prisoner movement, visitor access, maintenance work, technology and relationships between inmates and officials all create potential points of failure.
The issue has returned to public attention after 17 undertrials escaped from Thekkampatti prison in Tamil Nadu’s Theni district this week. According to a report by The Times of India, Delhi jail officials maintain that the capital’s three prison complexes already operate under stringent, multi-layered security arrangements. The complexes are Tihar, which can house 14,500 inmates, Rohini, with capacity for about 2,000, and Mandoli, with capacity for more than 4,000.
Together, the figures point to a prison system designed to manage more than 20,000 inmates across three separate sites. That scale matters because security is not limited to keeping people inside cells. It also involves controlling gates, watchtowers, high-security zones, internal circulation, court and medical movements, maintenance access, communications and the conduct of staff and visitors. Every additional movement or access point adds an administrative task that must be managed consistently.
Tihar’s history demonstrates how those risks have changed over time. The first unauthorised exit recorded in the report occurred in 1965, when American smuggler Daniel Walcott left the prison. In 1976, 13 prisoners escaped by digging a tunnel near the prison wall. That incident was followed by the deployment of Tamil Nadu Special Police at Tihar in August 1980. The stated rationale was to bring personnel from another state into the prison system, reducing the risk of local connections and possible collusion between prisoners and staff.
Other episodes exposed different weaknesses. In 1983, several students arrested for arson and rioting exploited the visitor identification system. They transferred visitor stamps from their wrists to their own during the May heat, allowing them to pass through a process that authorities failed to reconcile with the number of people entering and leaving. The episode was not primarily a failure of the prison wall; it was a failure of identity control and basic movement accounting.
In 1986, Charles Sobhraj, who was lodged in Tihar for allegedly committing 20 murders, escaped after drugging security guards on his birthday, according to the report. In 2004, Sher Singh Rana, the killer of former parliamentarian Phoolan Devi, escaped after an aide posed as a police officer to take him for a trial hearing. In 2011, two undertrial drug users escaped while being transported in a van with eight other inmates, although they were later caught. In 2013, a prisoner died after falling from a moving truck while trying to escape from inside the prison.
The 2015 episode again showed how escape risks can arise from the interaction between people and procedures. Two inmates fled; one was held, while the other escaped before being caught later. Sunil Kumar Gupta, a former spokesperson and legal adviser to Tihar Jail, told The Times of India that inmates in different cases had sought to meet their families or escape the restrictions and rules of prison life. His account also refers to a 1981 escape in which an inmate left on the pretext of receiving medical treatment and to the 2004 case in which an escaped inmate reportedly asked a jail official for directions to a bus stop while outside the prison.
These incidents point to a central feature of prison security: the system must manage both deliberate deception and routine institutional activity. A disguised visitor, a claimed medical need, a court appearance, a maintenance worker entering a restricted area or an inmate being moved in a vehicle can all become security-sensitive events. The operational challenge is to ensure that ordinary processes remain usable without allowing them to become informal exits.
Delhi’s current arrangement responds to those risks through multiple institutional layers. More than 5,000 personnel, including prison staff and external security forces, are deployed across the three complexes. The four-layer system places prison staff in the first layer, Tamil Nadu Special Police in the second, the Indo-Tibetan Border Police in the third and the Central Reserve Police Force in the fourth.
The division of responsibilities is designed to separate functions. Tihar staff manage internal prison operations. Tamil Nadu Special Police personnel guard watchtowers and high-security areas. The Indo-Tibetan Border Police handle day-to-day patrolling, while the Central Reserve Police Force guards the gates. The model reflects an attempt to avoid concentrating all responsibilities within one group and to create overlapping checks between internal administration and external security forces.
Technology provides another layer. More than 7,000 cameras are installed across the three prison complexes, with dedicated rooms for monitoring footage. Mobile signal jammers operate at Tihar and Mandoli, while Rohini also has jamming systems intended to restrict unauthorised mobile phone use. Authorities conduct routine security assessments, cell searches and surprise inspections to identify vulnerabilities and prohibited items.
However, the presence of cameras and jammers does not itself establish that a prison is secure. Surveillance depends on monitoring, response and the ability to act on what personnel observe. Signal jamming addresses unauthorised communication, but it does not resolve risks associated with impersonation, staff collusion, inadequate identity checks or prisoner movement. The historical record supplied in the report shows that escapes have used different methods, which is why a system built around only one technology or one physical perimeter would be insufficient.
Movement outside prison premises is treated as a separate security concern. Delhi Armed Police personnel escort inmates when they are taken outside. This includes the kinds of movements that have featured in earlier escape cases, such as trial hearings and claimed medical treatment. The 2004 escape involving a person posing as a police officer illustrates why authority credentials and escort procedures are as important as walls and gates.
Maintenance work is also incorporated into the security design. Inmates are not permitted to clean sewers. Instead, the relevant agency is called to perform the work, and its personnel must undergo security checks and enter and leave through a designated gate. This arrangement reduces the risk that prisoners can access underground infrastructure or that outside workers can move through the complex without a controlled record.
The policy choice has a practical trade-off. Bringing in external agencies can reduce the risks associated with prisoner access to sensitive work, but it also creates additional people, vehicles and movements that the prison administration must verify. The value of the arrangement therefore depends on whether entry and exit controls are consistently applied, not merely on the fact that a designated agency has been assigned the task.
Staff management forms a further part of the security framework. Prison personnel are periodically transferred to prevent overly close relationships with inmates and reduce the risk of collusion. This recognises that prison security is partly a relationship-management problem. The same staff members operate within a closed institution over long periods, while inmates depend on officials for access, movement and services. Institutional safeguards therefore have to account for familiarity as well as physical access.
The evidence in the report shows an evolution from responding to individual escape incidents to building a system of overlapping controls. The 1976 tunnel escape was followed by the introduction of Tamil Nadu Special Police. Later measures address surveillance, communications, searches, movement, maintenance access and staff transfers. This is a cumulative security architecture, shaped by past failures rather than by one comprehensive intervention.
The data also reveals the scale of the challenge. More than 20,000 inmates can be housed across the three complexes, while more than 5,000 personnel and over 7,000 cameras support security operations. Those numbers indicate substantial deployment, but they do not provide a direct measure of effectiveness. The report does not establish how many cameras are actively monitored at any given time, how quickly alerts are acted upon, or how often security assessments lead to specific corrective measures. Those remain important operational questions.
The larger urban issue is that prisons are not isolated from the systems around them. They depend on police escorts, courts, medical services, contractors, communications controls and public agencies carrying out specialised work. Their security is therefore a governance problem spanning multiple institutions. A weakness in one process can undermine safeguards installed elsewhere.
Tihar’s escape history does not show that current arrangements have failed, nor does the presence of multiple layers prove that every vulnerability has been eliminated. It does show why prison security must be treated as a continuously managed system. Delhi’s current model combines personnel from several forces, surveillance technology, restricted access, inspections and staff rotation. The next question for administrators is not simply whether more security exists, but whether each layer remains coordinated, monitored and tested against the changing methods by which inmates may attempt to leave.
