Paro: the $6,000 seal robot for elderly care, explained. Offbrand Human

Paro is the seal robot for elderly care that turns up in every search: a hand-made robotic baby harp seal, built at Japan's national research institute, sold to hospitals and care homes through country-by-country distributors, and used mostly in dementia units. The $6,000 everyone quotes for it is not a maker's price. It comes from the research literature, where reviewers cite it as an obstacle to access. The only distributor that publishes a number is Sense Medical in the UK, which lists Paro at 6,000 pounds excluding VAT, about $8,070 at the European Central Bank's 2026-08-07 reference rate of 1.345. In the United States there is no published price at all.

Paro is real, decades old, in serious clinical use, and sold the way an MRI coil is sold, by quote.

What Paro actually is

The design has been in service since 2003. It was developed at AIST, Japan's National Institute of Advanced Industrial Science and Technology, by Dr. Takanori Shibata. The name is short for Personal Robot, per the developer's own FAQ. Guinness World Records certified it as the most therapeutic robot in 2002.

It carries five kinds of sensor: tactile, light, audition, temperature and posture. It learns a name if you use one enough. It repeats behaviors that earned it stroking and drops behaviors that got it hit. It runs a daily rhythm, active in the afternoon and sleepy at night. The fur is antibacterial and the body carries electromagnetic shielding so it can be handled by someone with a pacemaker, and the developer says the tactile sensor was tested past 100,000 strokes.

Each one is made by hand, so each face is slightly different. The developer's FAQ page, which has not been updated since the mid 2000s, put the Japanese installed base at more than 1,000 units.

One detail for anyone comparing quotes: parorobots.com still describes the current product as the eighth generation, and that page carries a 2014 copyright and a Flash video plugin. The Australian trial that ran from June 2014 to May 2015 used Paro version 9.

What it costs, and who publishes a price

Four priced cards for the therapy-seal decision, lowest first: a Joy for All companion cat at $159.99 with no app, WiFi or subscription; the approximately $6,000 the research literature quotes for Paro, from JMIR Rehabil Assist Technol 2021, JMIR Aging 2022 and Front Psychiatry 2023 which gives about 6,000 euros per unit; Sense Medical in the UK at 6,000 pounds excluding VAT, about $8,070 at a rate of 1.345, the only published list price anywhere; and two US federal purchase records naming Paro, $18,162 in June 2012 through a reseller and $22,800 in September 2014 direct, neither stating how many seals were in the box

Channel	What it publishes
Sense Medical, UK and Ireland	6,000 pounds ex VAT (about $8,070)
Le Phoque PARO, France, Belgium, Luxembourg	No price. Quote only, with 24-month installment purchase or 36 and 48-month lease-to-own. A one-week trial costs 90 euros before tax (about $104) plus shipping both ways
PARO Robots U.S., Inc., Itasca, Illinois	No price, no store, sales by phone or email
Research literature	"approximately US $6000" (JMIR Rehabil Assist Technol, 2021), "about US $6000" (JMIR Aging, 2022), "approximately 6,000 euros per unit" (Front Psychiatry, 2023)

The nearest thing to an American receipt sits in federal spending records. The Department of Veterans Affairs bought therapeutic robots from PARO Robots US for $22,800 in September 2014 for a new polytrauma and blind rehabilitation center, and a VA extended care unit bought a Paro through a reseller for $18,162 in June 2012. Neither record says how many seals were in the box, so neither gives a unit price, and those two are the only federal awards that name Paro.

None of it is a subscription. Batteries are replaceable, and the French distributor puts the working life at around ten years with maintenance, a vendor estimate rather than a measured one.

The FDA line, stated precisely

Care facilities lean on Paro's FDA status to justify the spend, and the phrasing drifts into "FDA approved" or "FDA cleared." Here is what the databases actually contain.

PARO Robots U.S., Inc. of Itasca, Illinois holds device establishment registration 3009118691, currently active, and lists its product under code HCC, which FDA classifies as a biofeedback device under 21 CFR 882.5050, class II, neurology panel. The listing record carries a created date of 2012-06-27. So "class II device listed with FDA" is accurate.

Two things follow. FDA states on its own registration and listing database that "registration of a device establishment, assignment of a registration number, or listing of a medical device does not in any way denote approval of the establishment or its products by FDA." And a search of the 510(k) clearance database by device name and by applicant returns no Paro record, though plenty of other biofeedback products under the same code have clearances.

That absence has a lawful explanation rather than a scandalous one. The regulation itself exempts a prescription, battery-powered biofeedback device indicated for relaxation training and muscle reeducation from premarket notification, subject to the limits in 21 CFR 882.9. A robot that lists under that code and never files a 510(k) is behaving normally. It does mean no FDA reviewer assessed Paro before it went on sale, which is worth knowing when the word "approved" appears in a pitch.

Europe classifies it more lightly still: the French distributor's FAQ calls Paro a class I medical device.

Three-box explainer of Paro's US regulatory status: what exists is device establishment registration 3009118691 for PARO Robots U.S., Inc. of Itasca, Illinois, product code HCC, a biofeedback device under 21 CFR 882.5050, class II, neurology panel, listing record created 2012-06-27; what does not exist is any record in FDA's 510(k) clearance database by device name or applicant, an absence made lawful by the premarket notification exemption at 21 CFR 882.9; and FDA's own sentence that registration or listing does not in any way denote approval of the establishment or its products by FDA

What the biggest trial found

The largest randomized test of Paro remains the Queensland cluster trial published in JAMDA in 2017. Twenty-eight long-term care facilities, 415 residents aged 60 and over with a dementia diagnosis, three arms: Paro version 9, a look-alike plush toy that was a Paro with the robotic features switched off, and usual care. Sessions were individual and nonfacilitated, 15 minutes, three times a week for ten weeks.

Against the plush toy, Paro won on engagement: residents were more verbally engaged (3.61, P = .011) and much more visually engaged (13.06, P < .0001) on the coded video. Against usual care, Paro improved observed pleasure (1.12, P = .008) and observed agitation (3.33, P = .008). On the Cohen-Mansfield Agitation Inventory, the scale that care plans and clinical notes are actually written around, there was no difference between the groups.

The companion economic paper, published in JAMDA in 2018, is the one buyers should read. Over the ten weeks the Paro group cost $50.47 more per resident than usual care and the plush toy group $37.26 more (the trial ran in Australia; the paper reports dollars per resident). The incremental cost-effectiveness ratios came out at $13.01 per agitation point averted for Paro and $12.85 for the plush toy. The authors' conclusion: the plush toy offered marginally greater value for money, and both beat conventional psychosocial and sensory programs on cost.

Effect sizes across the wider robotic-pet category, including medication findings from a separate Paro trial in secure dementia units, sit in the therapy evidence roundup.

Who actually buys one

Facilities, not families. The distributor list on Paro's contact page runs to fifteen entries and reads like a medical supply directory: Cyberdyne Care Robotics in Germany, Picomed in Norway, the Danish Technological Institute for the rest of Europe, IDS Medical in Hong Kong and Singapore, Alpha Lifecare in Australia. Several run Paro certification training for clinicians, which is part of what the price buys.

CNBC filming Paro inside an elderly care centre, which is the buyer this whole product is built and priced for. The channel's own line for the place: Paro "is considered a member of staff".

Reimbursement is patchy and mostly not American. The French distributor points buyers at regional health agency project funding and philanthropy, and says Japan and Germany offer partial reimbursement. In the US, the VA records above are the visible public path, and there is no consumer program that hands a Paro to a family the way six states hand out ElliQ.

If you are shopping for one person

Start by asking, not buying. If your parent is in a care home, memory unit or hospice, call the activities director and ask whether the facility owns a Paro, who is trained on it, and how often residents get time with it. Ask whether that time is one to one: the trial results above came from individual sessions, not group activities.

If you are buying for someone at home, the seal is the wrong end of the market. A Joy for All companion cat is $159.99 with no app, no WiFi and no subscription, under 3 percent of the figure the literature quotes for Paro, and it is the tier that the buyer's guide for seniors points most families toward.

And keep the Queensland comparison in view when a brochure quotes the clinical literature at you. The thing Paro was measured against in its largest trial was itself with the electronics turned off, sitting in the same laps for the same fifteen minutes. The robot pulled ahead on attention and pleasure. Per dollar of agitation relief, the switched-off version finished level.